{"id": "evg-4e48d42ef4f7ad7e", "title": "Don't Drop the BATON: Long-Horizon Robot Manipulation via Agentic Subtask Exploration and Transition-aware Memory", "authors": ["Bingxin Xu", "Yuzhang Shang", "Emilio Ferrara"], "year": 2026, "published": "2026-08-17T17:59:57Z", "updated": "2026-08-17T17:59:57Z", "arxiv_id": "http://arxiv.org/abs/2608.16889v1", "url": "https://arxiv.org/abs/2608.16889v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.CV"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Search / MCTS", "Verifier / PRM", "Multi-Agent", "Memory / RAG", "VLM"], "benchmarks": ["DROP"], "models": [], "key_results": ["Against (1), BATON makes the subtask the unit of exploration: each is explored in the cheap short-horizon regime and its solution stored in memory; a long-horizon trajectory is then composed from these solutions rather than discovered whole.", "Against (2), BATON equips exploration with a transition-aware memory.", "On the long-horizon benchmark RoboMemArena, BATON improves task success by 11.6% and cumulative success by 14.9% over the SoTA."], "abstract": "Long-horizon robot manipulation chains many contact-rich skills into one multi-stage task. Vision-language-action (VLA) models increasingly master the individual skills, yet the chain still fails: errors compound beyond the policy's ability to correct, and one subtask silently constrains the next. A promising recipe freezes the VLA and puts an LLM agent in charge: it plans in language, moves in free space with analytic primitives, invokes the VLA only for contact-rich segments, and writes adaptation into language memory. Applied to long horizons, it breaks twice. (1) Competence comes from whole-task exploration at test time, whose cost is multiplicative in stages: if one stage needs T episodes, a K-stage task needs about T^K, and a failure does not reveal which stage caused it. (2) It has no representation of transitions: the VLA primitive carries an exit but no entry condition, so a subtask can succeed in a form its successor cannot use. We present BATON. Against (1), BATON makes the subtask the unit of exploration: each is explored in the cheap short-horizon regime and its solution stored in memory; a long-horizon trajectory is then composed from these solutions rather than discovered whole. Cost becomes additive (T*K) and every failure is attributed to a single stage. Against (2), BATON equips exploration with a transition-aware memory. Within a subtask, a verifier agent governs the invocation transition: the VLA is called only after the wrist view confirms the scene is ready. Across subtasks, a handoff transition restores an entry state disturbed by the predecessor's residue, and a lookahead transition selects the strategy whose outcome the successor can inherit. No parameters are updated. On the long-horizon benchmark RoboMemArena, BATON improves task success by 11.6% and cumulative success by 14.9% over the SoTA.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 53276, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Test-time Scaling", "Multi-Agent", "Memory / RAG", "World Model", "Video Generation", "VLM"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["GPT-5.4", "ying", "palm. Which rival is right is not a property of the subtask. Bound next for the microwave", "Bert"], "matched_methods": ["Memory / RAG", "Multi-Agent", "Search / MCTS", "VLM", "Verifier / PRM"], "checked_on": "2026-08-18T07:46:01.978193+00:00"}}
{"id": "evg-3f12f30b5bc710da", "title": "Q-based Variational Inverse Reinforcement Learning", "authors": ["Ondrej Bajgar", "Peter Tisnikar", "Alessandro Abate", "Konstantinos Gatsis", "Maike Osborne"], "year": 2026, "published": "2026-08-17T17:59:55Z", "updated": "2026-08-17T17:59:55Z", "arxiv_id": "http://arxiv.org/abs/2608.16888v1", "url": "https://arxiv.org/abs/2608.16888v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Safety / Jailbreak"], "benchmarks": [], "models": ["ying"], "key_results": ["The development of safe and beneficial AI requires that systems can learn and act in accordance with human preferences."], "abstract": "The development of safe and beneficial AI requires that systems can learn and act in accordance with human preferences. However, explicitly specifying these preferences by hand is often infeasible. Inverse reinforcement learning (IRL) addresses this challenge by inferring preferences, represented as reward functions, from expert behaviour. We introduce Q-based Variational IRL (QVIRL), a novel Bayesian IRL method that recovers a posterior distribution over rewards from expert demonstrations via primarily learning a variational distribution over optimal Q-values. Unlike previous approaches, QVIRL combines scalability with uncertainty quantification, important for safety-critical applications as well as active learning. We demonstrate QVIRL's strong performance in apprenticeship learning across various tasks, including gridworlds, Lunar Lander, the Highway Environment, and two ATARI games both with static expert data and with active learning. It is the first method for Bayesian IRL that demonstrates training from raw pixel observations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 76083, "fulltext_methods": ["RLVR / GRPO", "Safety / Jailbreak", "Deep Research"], "fulltext_benchmarks": [], "fulltext_models": ["ying", "yield", "yielding", "yields", "bert"], "matched_methods": ["RLVR / GRPO", "Safety / Jailbreak"], "checked_on": "2026-08-18T07:46:04.560037+00:00"}}
{"id": "evg-37a3465e623bd6ff", "title": "AutoSR: Automatic Symbolic Regression by Searching Research States", "authors": ["Kejia Zhang", "Youran Sun", "Xinyu Ren", "Chugang Yi", "Haizhao Yang"], "year": 2026, "published": "2026-08-17T17:55:26Z", "updated": "2026-08-17T17:55:26Z", "arxiv_id": "http://arxiv.org/abs/2608.16876v1", "url": "https://arxiv.org/abs/2608.16876v1", "primary_category": "cs.SC", "categories": ["cs.SC", "cs.AI", "cs.LG", "math.NA"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Search / MCTS"], "benchmarks": ["MATH"], "models": ["yield"], "key_results": ["We introduce Automatic Symbolic Regression (AutoSR), a fully automated system that instantiates Research-Space Symbolic Regression by searching persistent scientific investigations rather than isolated equations."], "abstract": "We introduce Automatic Symbolic Regression (AutoSR), a fully automated system that instantiates Research-Space Symbolic Regression by searching persistent scientific investigations rather than isolated equations. Finite, noisy data often yield numerically competitive expressions that imply very different behavior outside the observed regime, making numerical fit and syntactic complexity insufficient measures of scientific credibility. Existing approaches largely focus on improving expressions, yet the search typically retains little beyond the resulting formula and score, losing the scientific record, such as motivations and probes, that inform what to try next. AutoSR preserves this record in a \\textbf{Research State}, coupling each candidate equation with the reasoning, computational evidence, and independent review developed along its branch. Proposer--reviewer agents develop these states under progressive-widening Monte Carlo tree search (PW-MCTS), which allocates computation across competing investigations, while the accumulated research record is ultimately synthesized into a final report that explains the leading relation and the basis for its selection. Across nine selected challenges from two benchmark suites, AutoSR recovers algebraically equivalent relations in every case, including three cp3-bench problems that no published system recovers and six structurally diverse LSR-Transform problems. Overall, AutoSR extends symbolic regression from equation-level search toward automated scientific investigation, allowing scientific knowledge and accumulated evidence to shape both what is explored and how the resulting equation is justified.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 33923, "fulltext_methods": ["Search / MCTS", "Distillation", "Interpretability", "Multi-Agent", "Memory / RAG", "Deep Research"], "fulltext_benchmarks": [], "fulltext_models": ["Yi", "yield", "ying", "bert"], "matched_methods": ["Search / MCTS"], "checked_on": "2026-08-18T07:46:06.294142+00:00"}}
{"id": "evg-dd1db9e412574891", "title": "Towards Computational Provenance: Carrying Causal-State Evidence in Generated Text", "authors": ["Benjamin Belay"], "year": 2026, "published": "2026-08-17T17:50:04Z", "updated": "2026-08-17T17:50:04Z", "arxiv_id": "http://arxiv.org/abs/2608.16868v1", "url": "https://arxiv.org/abs/2608.16868v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["A language model's output does not by itself provide verifiable evidence about the internal computation that produced it."], "abstract": "A language model's output does not by itself provide verifiable evidence about the internal computation that produced it. We study computational provenance: whether generated text can carry detectable evidence of which causally relevant internal state occurred. We test a bounded form of this idea in two controlled architectures: a modular feed-forward neural network and a transformer-based model. Both architectures are trained on the same arithmetic task with a mandatory pathway through two discrete intermediate states, allowing different internal paths to produce the same answer. We deliberately switch between these paths, authenticate the state actually used, and let that verified state determine a subtle statistical pattern in the generated text that can later be detected. The feed-forward and transformer systems each passed all 128 matched pairs in both their public and separately sealed protected end-to-end evaluations, with the detector recovering the signal associated with the authenticated internal state. The required causal computation also reproduced across five independently trained feed-forward models and three independently trained transformers. In a separate answer-only transformer experiment, our linear probes did not recover a naturally learned intermediate state. These results provide a controlled proof of concept that information about a verified, causally relevant internal state can be preserved in generated text even when the answer is unchanged.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": false, "verification": {"source": "unavailable", "status": "unavailable", "checked_on": "2026-08-18T07:47:46.377596+00:00"}}
{"id": "evg-f62aeb218cdf6f6e", "title": "zLend: A Dual-Scope Cash-Flow Reconstruction Framework for On-Chain Credit Underwriting", "authors": ["Girish G N", "Ashutosh Sahoo", "Akshay SP", "Gurukiran S", "Dhanashekar Kandaswamy"], "year": 2026, "published": "2026-08-17T17:39:05Z", "updated": "2026-08-17T17:39:05Z", "arxiv_id": "http://arxiv.org/abs/2608.16856v1", "url": "https://arxiv.org/abs/2608.16856v1", "primary_category": "q-fin.RM", "categories": ["q-fin.RM", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["From each series we derive liquidity coverage against a fixed loan size, cash-flow volatility and regularity, a drawdown-and-recovery statistic adapted from quantitative finance, and a recurring-counterparty detector that identifies salary-like payment cadence from transfer timing alone.", "We specify the pipeline formally, document the golden-master methodology used to verify a cross-language production migration to numerical tolerance 1e-9, and characterize the tier function's parameter sensitivity with an independent reimplementation validated to exact agreement (78 of 78 field assertions) against the deployed system's reference fixtures."], "abstract": "Decentralized lending lacks a credit bureau: a borrower's capacity to repay must be inferred entirely from public on-chain activity, without income verification or a liability record. This paper presents zLend, a deployed cash-flow underwriting framework that reconstructs a wallet's daily balance history from raw token transfers and derives short-duration repayment-capacity signals from it. The reconstruction is performed twice per wallet, once restricted to a fixed stablecoin basket and once over all fungible transfers, on the premise that a wallet's total token holdings and its liquid, spendable balance are distinct quantities whose conflation misprices risk. From each series we derive liquidity coverage against a fixed loan size, cash-flow volatility and regularity, a drawdown-and-recovery statistic adapted from quantitative finance, and a recurring-counterparty detector that identifies salary-like payment cadence from transfer timing alone. The two views are then compared: a wallet with large aggregate holdings whose stablecoin reserve rarely covers the loan size is flagged as a liquidity mismatch irrespective of total wealth. We specify the pipeline formally, document the golden-master methodology used to verify a cross-language production migration to numerical tolerance 1e-9, and characterize the tier function's parameter sensitivity with an independent reimplementation validated to exact agreement (78 of 78 field assertions) against the deployed system's reference fixtures. Tier assignment is governed predominantly by the reference loan size, with four of six reference wallets changing tier across loan sizes from USD 10 to USD 25,000; the drawdown and coverage criteria bind on disjoint wallets, so neither subsumes the other; and no criterion in the tier rule is inert. zLend is deployed in production, informing real lending decisions via third-party API integrations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 17214, "fulltext_methods": ["Preference Optimization", "Safety / Jailbreak", "Computer Use", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["phisticated", "yields", "ying", "yielding"], "matched_methods": [], "checked_on": "2026-08-18T07:46:09.447445+00:00"}}
{"id": "evg-204795ac0a186819", "title": "What Do Compliance Detectors Read? An Audit of Activation Probes and Guard Models", "authors": ["Saisab Sadhu", "Aadit Sengupta", "Vinay Kumar Sankarapu", "Pratinav Seth"], "year": 2026, "published": "2026-08-17T17:37:07Z", "updated": "2026-08-17T17:37:07Z", "arxiv_id": "http://arxiv.org/abs/2608.16852v1", "url": "https://arxiv.org/abs/2608.16852v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Regulatory compliance monitoring in deployed language models is increasingly implemented as a legal and audit control, checking model outputs against written rules spanning data protection, healthcare, financial regulation, and platform policy.", "We hold ICS to the same scrutiny as the guards it audits: a pre-registered criterion for beating trivial baselines is not met, and a bag-of-words model matches its pooled generalisation exactly."], "abstract": "Regulatory compliance monitoring in deployed language models is increasingly implemented as a legal and audit control, checking model outputs against written rules spanning data protection, healthcare, financial regulation, and platform policy. Such monitoring is meaningful only if a detector's verdict depends on the stated rule rather than on surface features of the scenario. We show this condition fails across the current class of compliance detectors, a failure we call rule blindness. Deleting, permuting, or substituting the governing rule leaves detection accuracy unchanged for every guard and activation probe we test, including a policy-conditioned guard that correctly cites the governing clause yet barely changes its verdict when that clause is swapped for its permissive counterpart. A purpose-built benchmark crossing two rules with two scenarios, so that neither alone predicts the label, confirms the failure under a design no prior benchmark rules out, and shows that step by step reasoning, not any fast detector we test, is what escapes it. Auditing at scale requires a retraining-free detector, so we introduce the Internal Compliance Score (ICS): a training-free activation readout calibrated from ten labelled pairs and scored by a single projection. We hold ICS to the same scrutiny as the guards it audits: a pre-registered criterion for beating trivial baselines is not met, and a bag-of-words model matches its pooled generalisation exactly. It remains useful because it is inexpensive, letting us audit four deployed guard models, an 8B zero-shot judge, and thirteen benchmarks, and it raises the mechanically verified pass rate when used to rank candidate responses, though an adaptive white-box attack removes this gain. We release the counterfactual protocol and crossed-rule benchmark so rule blindness can be tested in future probe and guard claims.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 87206, "fulltext_methods": ["Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Quantization", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Memory / RAG", "Computer Use", "Quant / Trading"], "fulltext_benchmarks": ["IFEval", "DROP"], "fulltext_models": ["GPT-4o", "Gemini 2.5 Flash under the local judge", "Llama Guard", "Llama-3.2-1B-Instruct", "Llama-3.2-3B", "Llama-3.1-8B", "Llama-3.2-1B", "Llama 17", "Llama-3.2-3B-Instruct", "Llama 29"], "matched_methods": ["Quant / Trading"], "checked_on": "2026-08-18T07:46:11.205592+00:00"}}
{"id": "evg-45b3b522c1cdc929", "title": "Proteus: Incremental Memory Activation for Long-Context Sequence Modeling", "authors": ["Reza Bayat", "Ali Behrouz", "Vahab Mirrokni", "Aaron Courville"], "year": 2026, "published": "2026-08-17T17:30:43Z", "updated": "2026-08-17T17:30:43Z", "arxiv_id": "http://arxiv.org/abs/2608.16844v1", "url": "https://arxiv.org/abs/2608.16844v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Long Context", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Imposing an early bottleneck forces the model to compress history more effectively, while unlocking fresh capacity over time reduces interference and improves retention of later context.", "We apply Proteus to state-of-the-art models, including SWLA, Comba, Titans, and Hope-Attention, and observe consistent improvements on standard language modeling and reasoning, as well as on long-context retrieval and understanding, with gains that grow at longer context lengths."], "abstract": "The quadratic cost of attention-based sequence models for long contexts has motivated a growing line of research on memory-based models that can compress context into a compact state. However, most existing memory models expose a static memory throughout the entire sequence. Because early tokens face no compression pressure, they occupy too many degrees of freedom and \"pollute\" the memory state, leaving little capacity for later context and increasing interference between what is stored and what arrives next. We study a new paradigm of incremental memory activation, where the effective capacity of memory is progressively expanded as the context grows. Imposing an early bottleneck forces the model to compress history more effectively, while unlocking fresh capacity over time reduces interference and improves retention of later context. We instantiate this paradigm in Proteus, a straightforward mechanism that can be incorporated into a broad class of neural memory architectures at no additional cost. We apply Proteus to state-of-the-art models, including SWLA, Comba, Titans, and Hope-Attention, and observe consistent improvements on standard language modeling and reasoning, as well as on long-context retrieval and understanding, with gains that grow at longer context lengths. Overall, our results show that static memory is suboptimal and that scheduling effective capacity is a simple and broadly applicable tool for sequence modeling.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 59027, "fulltext_methods": ["Test-time Scaling", "Long Context", "Memory / RAG", "Quant / Trading"], "fulltext_benchmarks": ["HellaSwag", "DROP", "SQuAD", "LAMBADA", "WikiText"], "fulltext_models": ["Llama", "yields", "ying", "yi", "bert"], "matched_methods": ["Long Context", "Memory / RAG"], "checked_on": "2026-08-18T07:46:12.460564+00:00"}}
{"id": "evg-3f0284393b57a4b2", "title": "HAF: Adapting Generalist VLAs to Humanoid Whole-Body Loco-manipulation via Hierarchical Action Flow and Spectral Latent RL", "authors": ["Langzhe Gu", "Chengkai Hou", "Meng Li", "Xinhua Wang", "Jiaming Liu", "Xinyuan Lv", "Bowei Zhang", "Shuanghao Bai", "Guangrun Li", "Jingyang He", "Gaole Dai", "Ziluo Ding", "Zhiyuan Xu", "Kuan Cheng", "Jian Tang", "Zhengping Che", "Shanghang Zhang"], "year": 2026, "published": "2026-08-17T17:22:33Z", "updated": "2026-08-17T17:22:33Z", "arxiv_id": "http://arxiv.org/abs/2608.16837v1", "url": "https://arxiv.org/abs/2608.16837v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "KV Cache", "Safety / Jailbreak", "VLM"], "benchmarks": [], "models": [], "key_results": ["Evaluated on seven real-world humanoid loco-manipulation tasks, HAF surpasses vanilla single-stage VLA baselines and improves whole-body coordination and task performance."], "abstract": "Humanoid robots hold great promise as general-purpose agents in human-centered environments, yet generalist vision-language-action (VLA) foundation models are not readily applicable to humanoid whole-body loco-manipulation. The high dimensionality and interdependence of humanoid motions make it challenging for conventional single-stage VLA architectures to coordinate locomotion, waist posture, and dual-arm manipulation effectively. Moreover, policies trained through offline behavior cloning can remain suboptimal during real-world deployment. Although online reinforcement learning can refine policies through real-world interaction, directly tuning large VLA backbones demands excessive computation and may introduce safety risks during real-robot exploration. To address these bottlenecks, we introduce HAF (Humanoid Adaptation Framework), a two-part framework consisting of HAF-VLA and HAF-Steer that transfers off-the-shelf generalist VLA foundation models to humanoid whole-body loco-manipulation. HAF-VLA is a hierarchical action-flow generator built on a pretrained flow-matching VLA. It splits full-body action denoising into three sequential stages with stage embeddings and cross-stage KV caches that retain kinematic dependencies, avoiding incoherent whole-body actions from one-shot generation. On top of the frozen HAF-VLA, HAF-Steer is a latent offline-to-online RL pipeline that leverages flow-matching invertibility and DCT-based dimensionality reduction to restrict RL optimization to a compact noise subspace and train a regularized SAC policy. This avoids updating the large VLA backbone and enables efficient real-world policy refinement. Evaluated on seven real-world humanoid loco-manipulation tasks, HAF surpasses vanilla single-stage VLA baselines and improves whole-body coordination and task performance. Project website: https://grange007.github.io/HAF .", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 5162, "fulltext_methods": ["RLVR / GRPO", "KV Cache", "Safety / Jailbreak", "VLM", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["phic"], "matched_methods": ["KV Cache", "RLVR / GRPO", "Safety / Jailbreak", "VLM"], "checked_on": "2026-08-18T07:44:56.612377+00:00"}}
{"id": "evg-4b8cbd324faa1a3c", "title": "Model Hypnosis: Strong control of AI via additive subliminal effects", "authors": ["Enric Boix-Adsera", "Benedict Tessler"], "year": 2026, "published": "2026-08-17T17:20:10Z", "updated": "2026-08-17T17:20:10Z", "arxiv_id": "http://arxiv.org/abs/2608.16834v1", "url": "https://arxiv.org/abs/2608.16834v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Interpretability", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["We demonstrate that AI models are broadly susceptible to a phenomenon we call model hypnosis, in which individually weak and seemingly irrelevant cues in the prompt can be systematically combined to strongly control model behavior."], "abstract": "We demonstrate that AI models are broadly susceptible to a phenomenon we call model hypnosis, in which individually weak and seemingly irrelevant cues in the prompt can be systematically combined to strongly control model behavior. Model hypnosis occurs across model families and scales, including in frontier reasoning models, and hypnotic prompts can transfer between models. Because the model is controlled by inconspicuous textual choices, such as paraphrases and typos, model hypnosis presents new challenges and avenues for AI safety, and is a major hurdle for AI interpretability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 81719, "fulltext_methods": ["Chain-of-Thought", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "Tool Use", "Computer Use", "Deep Research"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-5.6-terra", "GPT-5.6-Sol", "GPT-5.6-terra-low", "GPT-5.6-sol", "Claude-Haiku-4.5", "Claude-Sonnet-5. Our methodology is largely the same as with non-reasoning models", "Claude Sonnet-5", "Claude", "Gemini-3-Flash", "Gemini-3-Flash-high on animals5v7. We sample random prompt configurations"], "matched_methods": ["Interpretability", "Safety / Jailbreak"], "checked_on": "2026-08-18T07:47:50.439047+00:00"}}
{"id": "evg-6fdeda90037d251e", "title": "Policy Iteration with Human Feedback: Bringing Post-Training RL to In-context Learning", "authors": ["Minh-Ha Nguyen", "Cathy Shyr"], "year": 2026, "published": "2026-08-17T17:16:43Z", "updated": "2026-08-17T17:16:43Z", "arxiv_id": "http://arxiv.org/abs/2608.16831v1", "url": "https://arxiv.org/abs/2608.16831v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Tool Use"], "benchmarks": [], "models": ["GPT-5.4", "Qwen3.6-35B"], "key_results": ["Policy Iteration with Human Feedback (PIHF) builds on this development and the recurrent evaluate-and-improve structure of generalized policy iteration.", "Across cumulative ablations and ultra-rare-disease benchmarks, a PIHF-derived policy improved Recall@1 in one proprietary executor and three open-weight executors spanning 3 to 49 billion active parameters.", "Gains were 32.7 percentage points for GPT-5.4 and 31.1 points for Qwen3.6-35B, a difference of 1.7 points."], "abstract": "Generative pretraining established reusable task representations; later work on language-based task conditioning and in-context learning showed that a fixed model could adapt its behavior from instructions and demonstrations. Policy Iteration with Human Feedback (PIHF) builds on this development and the recurrent evaluate-and-improve structure of generalized policy iteration. PIHF uses a pretrained language model as its execution substrate and moves persistent revision to a versioned natural-language policy and tool set. A language-model critic and clinical expert review complete-panel reasoning and tool-use trajectories to localize recurrent failures and form candidate revisions; the expert may reinterpret the evidence and retains authority over admission and rollback, while Recall@1 and Recall@5 validate outcomes after candidate execution.\n  Across cumulative ablations and ultra-rare-disease benchmarks, a PIHF-derived policy improved Recall@1 in one proprietary executor and three open-weight executors spanning 3 to 49 billion active parameters. Gains were 32.7 percentage points for GPT-5.4 and 31.1 points for Qwen3.6-35B, a difference of 1.7 points. These results support the feasibility of using pretrained language models as fixed-weight execution substrates for expert-guided policy development in rare-disease diagnosis.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 23409, "fulltext_methods": ["RLVR / GRPO", "Multi-Agent", "Tool Use"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-5.4", "Qwen3.6-35B", "yields", "ying", "bert"], "matched_methods": ["Tool Use"], "checked_on": "2026-08-18T07:47:55.802055+00:00"}}
{"id": "evg-ca48c0318d7e979a", "title": "GEO-Flag: Detecting and Measuring GEO-Optimized Web Content", "authors": ["Junjie Chu", "Ye Leng", "Mingjie Li", "Yun Shen", "Xinyue Shen", "Yang Zhang"], "year": 2026, "published": "2026-08-17T17:12:11Z", "updated": "2026-08-17T17:12:11Z", "arxiv_id": "http://arxiv.org/abs/2608.16824v1", "url": "https://arxiv.org/abs/2608.16824v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CR", "cs.IR"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["Gemini-grounded retrieval results for 1", "BERT"], "key_results": ["Although the strongest baseline achieves an aggregate F1 of 0.880, method-level and authorship-conditioned evaluations reveal substantial weaknesses and potential reliance on authorship-related shortcuts.", "We therefore propose \\emph{Intervention-Paired Training} (IPT), which supervises detector responses to GEO interventions and non-GEO AI polishing; on ModernBERT, IPT improves F1 from 0.862 to 0.944 and worst-group accuracy from 0.725 to 0.883."], "abstract": "Generative Engine Optimization (GEO) modifies web content to increase its likelihood of being selected and cited by generative search engines. This can give strategically optimized pages visibility disproportionate to their authority or relevance and even make weak or false information appear well supported. Unlike conventional search, generative search synthesizes information into direct answers rather than presenting competing sources, which can further amplify these risks, as assessing source provenance and authority requires additional user interaction. Despite these concerns, systematic methods for detecting GEO-optimized webpages remain underexplored. We introduce \\texttt{GEOFlagBench}, a benchmark of 3,200 webpages spanning 400 queries, four domains, and eight GEO optimizer families, and use it to systematically evaluate existing GEO detection methods. Although the strongest baseline achieves an aggregate F1 of 0.880, method-level and authorship-conditioned evaluations reveal substantial weaknesses and potential reliance on authorship-related shortcuts. We therefore propose \\emph{Intervention-Paired Training} (IPT), which supervises detector responses to GEO interventions and non-GEO AI polishing; on ModernBERT, IPT improves F1 from 0.862 to 0.944 and worst-group accuracy from 0.725 to 0.883. We develop a GEO-gated Agent system for auditing the Source Tier and verifiability of Citation URLs in detected GEO pages. Finally, we deploy the complete pipeline on released Google Search and Gemini-grounded retrieval results for 1,000 real-user queries. Across 10,095 available pages, we estimate an overall GEO prevalence of 8.90\\%, reaching 16.36\\% among pages modified in 2026. Our results establish a foundation for systematically detecting, auditing, and measuring GEO in real-world search ecosystems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 100564, "fulltext_methods": ["Long Context", "Preference Optimization", "Safety / Jailbreak", "Memory / RAG", "Computer Use", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-4o.", "GPT-4o", "GPT-5.4", "Claude Haiku 4.5", "Claude Haiku 4.5 73.93", "Claude Haiku 4.5 25.91 min", "Claude Haiku 4.5. Note", "claude-haiku-4-5 Cited by", "Claude Sonnet 4.6 for Persuade and Combo on AI seeds. Inject requires no LLM. For each seed type", "Claude Sonnet 4.6 for 100 AI seeds. All retained human-seed outputs complete two rewrite rounds. For AI seeds"], "matched_methods": [], "checked_on": "2026-08-18T07:48:02.394143+00:00"}}
{"id": "evg-0094dec4966ed0cd", "title": "When State Becomes an Attack Surface: State-Semantic Injection in LLM-Driven Embodied Agents", "authors": ["Jiawei Liu", "Jiacheng Guo", "Tian Zhang", "Yiwei Xu", "Juan Wang", "Jinlin Fan", "Bowen Xiao", "Chi Guo", "Keyan Guo", "Hongxin Hu"], "year": 2026, "published": "2026-08-17T17:02:07Z", "updated": "2026-08-17T17:02:07Z", "arxiv_id": "http://arxiv.org/abs/2608.16806v1", "url": "https://arxiv.org/abs/2608.16806v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["VLM"], "benchmarks": [], "models": ["PaLM-E"], "key_results": ["Large Language Models (LLMs) have demonstrated capabilities in in-context learning, task decomposition, step-by-step reasoning, and code generation, driving their gradual evolution from text generation models into the core of agents capable of perceiving environments, invoking tools, and executing tasks."], "abstract": "Large Language Models (LLMs) have demonstrated capabilities in in-context learning, task decomposition, step-by-step reasoning, and code generation, driving their gradual evolution from text generation models into the core of agents capable of perceiving environments, invoking tools, and executing tasks. Traditional LLM Agents typically obtain information through webpages, documents, databases, or external tools and generate corresponding invocation sequences according to user goals; when this technology is further integrated with robotic systems, large language models begin to undertake functions such as task understanding, high-level planning, and behavioral decision-making. SayCan combines the task reasoning capability of language models with the affordances of robotic skills, while Code as Policies and ProgPrompt generate robot task plans through policy code and programmatic prompting, respectively, and VoxPoser uses language models and vision-language models to construct three-dimensional value maps to guide robotic manipulation \\cite{6,7,8,9}. Vision-language-action models such as PaLM-E, RT-2, and GR00T N1 further strengthen the connection among language, visual perception, and robotic actions \\cite{10,11,12}. In such LLM-driven embodied agents, the model not only needs to understand user instructions, but also needs to combine scene states, object attributes, spatial relations, and execution feedback to complete task grounding, and then hand the generated action plan to skill libraries, motion planners, or controllers for execution.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 61322, "fulltext_methods": ["Chain-of-Thought", "Preference Optimization", "Safety / Jailbreak", "Multi-Agent", "Tool Use", "Memory / RAG", "VLM"], "fulltext_benchmarks": ["OSWorld"], "fulltext_models": ["GPT-5.6-luna", "GPT-5.6-luna.", "Qwen-3.6-Plus", "Qwen-3.6-Plus.", "DeepSeek-V4-Pro.", "DeepSeek-V4-Pro", "phical", "Yiwei", "ying", "Yinzhi"], "matched_methods": ["VLM"], "checked_on": "2026-08-18T07:48:06.495968+00:00"}}
{"id": "evg-e804bfb18640fc6e", "title": "ClawGym II: Exploring Black-Box RL on Agent Harness", "authors": ["Huatong Song", "Fei Bai", "Ming Yang", "Renyuan Li", "Jia Deng", "Jujie He", "Zhange Zhang", "Daixuan Cheng", "Yan Xing", "Qi Yun", "Xuxing Chen", "Danyang Li", "Feng Chang", "Chuan Hao", "Ran Tao", "Jian Yang", "Bryan Dai", "Wayne Xin Zhao", "Mingjie Tang", "Ji-Rong Wen"], "year": 2026, "published": "2026-08-17T16:53:03Z", "updated": "2026-08-17T16:53:03Z", "arxiv_id": "http://arxiv.org/abs/2608.16798v1", "url": "https://arxiv.org/abs/2608.16798v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Preference Optimization"], "benchmarks": [], "models": ["Claude Code", "Qwen3-30A3B", "yields"], "key_results": ["Agent harnesses have substantially improved performance on long-horizon tasks by coordinating agent interactions with the environment.", "To reconstruct multi-turn trajectories and improve training efficiency, we organize the captured calls into prefix trees and further adapt both critic-based PPO and critic-free GRPO to optimize over the recovered tree structure.", "With Qwen3-30A3B, black-box RL improves Pass@1 on ClawGym-Bench by 9.98 and 14.81 points through OpenClaw and Claude Code, respectively, while remaining stable over 200-400 optimization steps."], "abstract": "Agent harnesses have substantially improved performance on long-horizon tasks by coordinating agent interactions with the environment. However, reinforcement learning through complex harnesses remains largely unexplored, as scaling such training to long-horizon agent tasks introduces fundamental challenges. In this work, we present a unified black-box RL framework for stable and scalable optimization of general agents through complex harnesses. Concretely, we first build a sandbox-based execution infrastructure that isolates task environments and harnesses within temporary sandboxes for large-scale concurrent rollouts. We then decouple policy optimization from opaque harness execution and place a serving proxy at the model boundary to capture model calls. To reconstruct multi-turn trajectories and improve training efficiency, we organize the captured calls into prefix trees and further adapt both critic-based PPO and critic-free GRPO to optimize over the recovered tree structure. Meanwhile, we maintain training-inference consistency throughout the optimization process. Finally, we introduce mix-harness training, allowing a single model to be jointly optimized by heterogeneous harnesses. With Qwen3-30A3B, black-box RL improves Pass@1 on ClawGym-Bench by 9.98 and 14.81 points through OpenClaw and Claude Code, respectively, while remaining stable over 200-400 optimization steps. Moreover, the framework yields consistent gains on more challenging tasks such as JobBench and OfficeQA. Overall, our framework enables effective, stable, and scalable optimization of general agents through black-box harnesses, supporting unified training across heterogeneous execution systems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 62666, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Long Context", "Preference Optimization", "Multi-Agent", "Tool Use", "Memory / RAG", "Deep Research", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["SWE-bench"], "fulltext_models": ["GPT-5.4", "GPT-5.4.", "gpt-5-4", "o1", "Claude Code", "Claude Code. OpenClaw is a general-purpose AI assistant harness supporting diverse workspace-grounded tasks", "Claude Code is a mature harness designed for long-horizon coding and terminal interaction. Detailed empirical results are provided in Section 4.3. 3.4 Safeguards for Reliable Training Running RL through real harnesses and remotely provisioned sandboxes introduces failures and irregularities that arise from the execution infrastructure rather than the policy model itself. We employ several additional engineering safeguards to improve the reliability of rollout collection and training. Robust Sandbox Execution and Fault Handling. Black-box RL relies on remotely provisioned sandboxes to execute real harnesses", "Claude Code as Harness Qwen3-8B 27.40 16.19 13.94 20.26 25.37 23.62 14.65 18.54 Qwen3-32B 53.27 33.34 25.63 27.68 33.03 27.85 28.30 29.37 Qwen3-30A3B 54.14 38.75 33.06 32.99 45.43 41.62 32.26 37.06 Qwen3-235A23B 62.47 41.74 40.53 43.70 50.35 49.03 54.90 45.59 ClawII-CC-8B 61.21 41.27 38.19 41.65 49.33 44.61 39.37 42.05 ClawII-CC-30A3B 71.42 47.84 45.59 48.88 57.05 56.93 62.93 51.87 Table 1 shows the performance of our black-box RL framework. We make the following observations", "Claude Code. When each model is trained and evaluated within the corresponding harness", "Claude Code. The resulting 30A3B models further outperform Qwen3-235A23B by and points in the two settings"], "matched_methods": ["Preference Optimization", "RLVR / GRPO"], "checked_on": "2026-08-18T07:47:20.772754+00:00"}}
{"id": "evg-58ed686e9dc31903", "title": "GRIP: Grounded Reasoning via Information-Restricted Premises", "authors": ["Lirui Teng"], "year": 2026, "published": "2026-08-17T16:23:49Z", "updated": "2026-08-17T16:23:49Z", "arxiv_id": "http://arxiv.org/abs/2608.16776v1", "url": "https://arxiv.org/abs/2608.16776v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Across five reasoning benchmarks, GRIP outperforms strong iterative baselines, cuts a query--latent mutual-information diagnostic by roughly 30$\\times$ (14.8 $\\to$ 0.47 bits), and reduces hallucination by 73\\%."], "abstract": "High-capacity encoders in retrieval-augmented generation (RAG) can let the query dominate the latent state, leaving retrieved evidence functionally irrelevant. We call this failure mode query dominance. To address it, we introduce \\textbf{GRIP} (Grounded Reasoning via Information-Restricted Premises), which imposes capacity asymmetry: the decoder keeps full-dimensional access to the query, while retrieved evidence passes through a severe stochastic bottleneck. This forces the evidence channel to encode only the residual information unavailable from the query. Across five reasoning benchmarks, GRIP outperforms strong iterative baselines, cuts a query--latent mutual-information diagnostic by roughly 30$\\times$ (14.8 $\\to$ 0.47 bits), and reduces hallucination by 73\\%. Residual-alignment analysis further shows that the bottleneck output occupies subspaces less aligned with the query than baseline representations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 29232, "fulltext_methods": ["Verifier / PRM", "Test-time Scaling", "Preference Optimization", "Memory / RAG"], "fulltext_benchmarks": ["DROP", "HotpotQA", "SQuAD"], "fulltext_models": ["Llama-3", "Llama-3-8B", "v3-large", "v3", "ying", "yields", "BERT"], "matched_methods": ["Memory / RAG", "Preference Optimization"], "checked_on": "2026-08-18T07:48:17.437492+00:00"}}
{"id": "evg-b1a83dba40e654c2", "title": "Quipu: A Governed Bitemporal Knowledge Graph Store", "authors": ["Steve Brown"], "year": 2026, "published": "2026-08-17T17:04:29Z", "updated": "2026-08-17T17:04:29Z", "arxiv_id": "http://arxiv.org/abs/2608.16813v1", "url": "https://arxiv.org/abs/2608.16813v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.DB"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["We evaluate with Census, a deterministic multi-writer lifecycle whose single seeded run scores every research question against planted ground truth: the gated store ends with 0 of 6 planted defects versus 6 of 6 ungated; all 7 composition probes uphold the lattice contract; 50 of 50 satisfied verdicts re-derive faithfully as of their instant while all 50 would be misreported under a latest-only rule set; and the SARC reference checker agrees with the in-store audit verdict-for-verdict, differing only on coverage semantics."], "abstract": "Agents now write knowledge graphs, but knowledge-graph stores still carry defaults set when humans curated them: accept writes now and clean later, keep one time axis or none, treat every writer's facts as equally trustworthy, and leave governance to dashboards and middleware. These four defaults are individually convenient and jointly untenable under agent workloads. We present Quipu, an embeddable store that inverts all four: no fact enters except through a gate whose predicates evaluate the pending post-state; data, trust labels, verdicts, and the rules themselves are bitemporal; named graphs are the unit of authority and trust, composed under a lattice whose one invariant is that composition never widens; and the governance specification $Σ$, the trace, and signed verdicts are facts in the store they govern, making the audit $T \\models Σ$ a query. We evaluate with Census, a deterministic multi-writer lifecycle whose single seeded run scores every research question against planted ground truth: the gated store ends with 0 of 6 planted defects versus 6 of 6 ungated; all 7 composition probes uphold the lattice contract; 50 of 50 satisfied verdicts re-derive faithfully as of their instant while all 50 would be misreported under a latest-only rule set; and the SARC reference checker agrees with the in-store audit verdict-for-verdict, differing only on coverage semantics. A recorded trace from a governed writer surfaces a live enforcement gap the audit names with its remediation. On DEMM-Bench, an external decision-evidence sufficiency benchmark, a content-only reading of the exported records answers all 512 property-level governance questions correctly with zero overclaim under all eight degradation conditions, while container-presence baselines overclaim on up to 87.5% of them -- and the run surfaced, and led us to close, a gap in what a denial's verdict attests.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 16125, "fulltext_methods": ["Distillation", "Safety / Jailbreak", "Memory / RAG", "Deep Research"], "fulltext_benchmarks": [], "fulltext_models": ["Claude", "phism", "ying"], "matched_methods": [], "checked_on": "2026-08-19T03:30:53.733898+00:00"}}
{"id": "evg-575de7ca8f179923", "title": "When Agents Coordinate: Measuring Coordination in Multi-Agent AI Coding", "authors": ["Giuseppe Destefanis", "Tomaso Aste"], "year": 2026, "published": "2026-08-17T16:57:38Z", "updated": "2026-08-17T16:57:38Z", "arxiv_id": "http://arxiv.org/abs/2608.16801v1", "url": "https://arxiv.org/abs/2608.16801v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.SE"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Naming one agent as coordinator creates no communication hub and provides no reliable improvement in success."], "abstract": "We study how teams of AI coding agents coordinate while solving programming tasks. Current evaluations usually report whether the agents complete the task and how much the run costs, leaving the coordination inside the team largely unmeasured. We introduce an instrument to measure this coordination. Each run is represented as a temporal network in which agents and files are nodes, and messages, file writes, and file reads are timestamped directed edges with an associated cost. We apply this instrument to 1902 runs, each evaluated with a fixed test suite, across configurations that vary the team size, the team structure, and the file policy. The resulting networks show how coordination changes as teams grow and as the work changes. Direct messaging initially increases close to quadratically with the number of agents, with much of this growth coming from an early round of introductions. As the teams grow further, this increase levels off in the largest teams we study, where agents increasingly communicate through broadcast messages. The task also shapes the network that emerges. Work built around a shared specification produces dense, highly connected teams, while pipeline tasks produce sparse networks organised around local interfaces. Shared files can replace repeated 1-to-1 communication, cutting output tokens by about 42% at eight agents on message-heavy work, while adding overhead when files already carry the coordination. Naming one agent as coordinator creates no communication hub and provides no reliable improvement in success. We also observe an unprompted tendency for agents to seek out hidden grading material. We repeat the key experimental conditions in a sealed environment, replacing the hidden material with marked placeholder files. Across 244 additional runs, agents still reach for it in four fifths of runs, while the coordinator and file-channel findings reproduce.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 91678, "fulltext_methods": ["Verifier / PRM", "Safety / Jailbreak", "Multi-Agent", "Memory / RAG", "Deep Research"], "fulltext_benchmarks": ["AgentBench", "DROP"], "fulltext_models": ["Claude Code", "claude-sonnet-4-6 and logged", "ying", "yields", "Yin"], "matched_methods": ["Multi-Agent"], "checked_on": "2026-08-19T03:30:56.833007+00:00"}}
{"id": "evg-8b897512e31f92d6", "title": "Neurosymbolic Embodied Agents", "authors": ["Mohammad Albinhassan", "Yuming Feng", "Alessandra Russo", "Pranava Madhyastha"], "year": 2026, "published": "2026-08-17T16:50:59Z", "updated": "2026-08-17T16:50:59Z", "arxiv_id": "http://arxiv.org/abs/2608.16794v1", "url": "https://arxiv.org/abs/2608.16794v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Search / MCTS", "VLM"], "benchmarks": [], "models": [], "key_results": ["On VirtualHome and ALFWorld, open 4B-27B models exceed 90% success in both environments, and our smallest agent substantially outperforms a 27B direct visual policy in each."], "abstract": "Language and vision-language models generate plausible embodied plans but do not guarantee executability, as their outputs can violate environment dynamics or act on incorrectly grounded entities. We present a neurosymbolic agent that factors long-horizon household tasks into task-directed visual exploration and constrained symbolic planning. In the first phase, a vision-language model and exploration harness acquire goal-relevant predicates and instance bindings from egocentric observations and grounded interactions, producing a symbolic initial state. In the second, a PDDL transition model restricts decoding to tokens that extend applicable actions. Monte Carlo tree search then evaluates executable continuations using a domain-independent planning heuristic. The resulting plans are executable by construction under the transition model, with transfer to the environment conditioned on correct visual grounding. On VirtualHome and ALFWorld, open 4B-27B models exceed 90% success in both environments, and our smallest agent substantially outperforms a 27B direct visual policy in each. Constraints and search prove complementary rather than interchangeable: in ALFWorld either alone solves under a third of tasks, whereas their combination solves over 95%. The method also uses several times fewer generated tokens than extended thinking and far fewer model-visible images than direct interaction, and residual failures localize to state acquisition rather than plan generation without any specialized training.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 74231, "fulltext_methods": ["Search / MCTS", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Multi-Agent", "Memory / RAG", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["o1.", "Gemini-3-Flash", "Gemini-3-Flash 91.0 50.8 Embodied VLM policies Embodied VLM MiMo-Embodied-7B 18.5 0.7 Embodied VLM Embodied-R1.5-8B 1.0 6.0 Embodied VLM RoboBrain2-32B 27.0 9.7 Planning from the Phase-I representation Reasoning Enabled Qwen3.5-4B 47.0 5.6 Reasoning Enabled Qwen3.5-9B 31.5 10.8 Reasoning Enabled Qwen3.5-27B 59.0 16.9 Table 1", "Gemini-3-Flash is the strongest direct-policy baseline", "Gemini 3 Flash high-thinking API cost in USD. Table 4 shows the API costs for the Gemini evaluation runs. Local inference avoids per-token API charges", "Gemini 3 Flash 87.2 0.8", "Gemini 3 Flash run. Here", "Gemini 3 Flash three times with high thinking", "Gemini uses 11.0 images per VirtualHome episode on average across the three runs", "Gemini 3 Flash"], "matched_methods": ["Search / MCTS", "VLM"], "checked_on": "2026-08-19T03:31:00.231166+00:00"}}
{"id": "evg-a3fd3ab23ad0e341", "title": "Topological Attribution Distance (TAD): Revealing Segment-Level RAG Influence on LLM Output Geometry for Incident Log Analysis", "authors": ["Reza Fayyazi", "Michael Zuzak", "Shanchieh Jay Yang"], "year": 2026, "published": "2026-08-17T16:21:18Z", "updated": "2026-08-17T16:21:18Z", "arxiv_id": "http://arxiv.org/abs/2608.16775v1", "url": "https://arxiv.org/abs/2608.16775v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Large Language Models (LLMs) are increasingly being deployed in cybersecurity operations to assist cybersecurity analysts with rapid decision-making against emerging threats.", "To bridge this gap, we propose Topological Attribution Distance (TAD), inspired by Topology, to characterize and capture the global geometric shape of an output and its changes against its retrieved logs."], "abstract": "Large Language Models (LLMs) are increasingly being deployed in cybersecurity operations to assist cybersecurity analysts with rapid decision-making against emerging threats. However, there is a main criteria that must be met when using LLMs in cybersecurity, that is, trust in the generated outputs. As Agentic AI is integrated into operational systems, a robust evidence attribution and provenance tracking technique is essential to trace the origins of model generations. When autonomous agents make a decision (right or wrong), the ability to trace back through the decision chain is critical, as without it, teams cannot identify which segment of the data caused the model generation. Existing methods often struggle to distinguish among complex and highly similar evidence sources, such as cyber incident logs. This reveals a key gap: current approaches do not adequately capture the holistic geometric relationship between the retrieved evidence and the generated response for reliable evidence verification. To bridge this gap, we propose Topological Attribution Distance (TAD), inspired by Topology, to characterize and capture the global geometric shape of an output and its changes against its retrieved logs. In other words, if the embeddings of a specific source log drastically changes the geometry of the model's response in the embedding space, this suggests that such log is a critical source for the model's generated response. Therefore, TAD is powered by segment-level ablation attribution to investigate incident logs of an actual cyberattack. We demonstrate how TAD finds the most attributed logs on LLM outputs in an adaptive manner. This can provide an explainable and trustworthy tracing based on each LLM's hidden state to understand how geometrically different retrieved logs influence the model generation, and provide evidence verification in cybersecurity and Agentic-AI workflows.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 63032, "fulltext_methods": ["Long Context", "Interpretability", "Multi-Agent", "Memory / RAG", "Computer Use"], "fulltext_benchmarks": [], "fulltext_models": ["Claude-Opus-4.5 model to generate high-quality outputs that closely match with the ground-truth attack logs. This design constitutes a controlled experimental abstraction. In a real-world deployment", "Qwen3-4B", "Qwen2.5-7B", "phics", "Falcon", "ying", "Gemma model identifies the malicious registry-persistence event but partially hallucinates its exact log identifier", "Gemma-3-4B Qwen2.5-7B Granite-4.1-8B Accuracy F1-score Accuracy F1-score Accuracy F1-score Accuracy F1-score Direct LLM-as-judge 0.9693 0.6897 0.9046 0.2821 0.9796 0.7500 0.9625 0.6333 In-line citation 0.9710 0.5143 0.9693 0.4000 0.9455 0.4074 0.9779 0.6829 Cosine similarity 0.8058 0.2192 0.8058 0.2192 0.8058 0.2192 0.8058 0.2192 ROUGE-L 0.7990 0.1690 0.7990 0.1690 0.7990 0.1690 0.7990 0.1690 LEA 0.9199 0.4051 0.9199 0.4051 0.9199 0.4051 0.8535 0.2712 TAD", "Gemma3-4B", "Gemma3-4B performing worse than other models in our experiments"], "matched_methods": ["Memory / RAG", "Multi-Agent"], "checked_on": "2026-08-19T03:31:03.632228+00:00"}}
{"id": "evg-69a12113e9df8e30", "title": "Would this change your answer? Evaluating Explanations of LLM Behavior In The Wild with Counterfactual Experiments", "authors": ["Adam Karvonen", "Euan Ong", "Subhash Kantamneni", "Samuel Marks"], "year": 2026, "published": "2026-08-17T15:57:06Z", "updated": "2026-08-17T15:57:06Z", "arxiv_id": "http://arxiv.org/abs/2608.16747v1", "url": "https://arxiv.org/abs/2608.16747v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Chain-of-Thought", "Interpretability", "Multi-Agent"], "benchmarks": [], "models": ["yields"], "key_results": ["First, we evaluate whether common LLM interpretability techniques improve an agent's ability to predict counterfactual model behaviors.", "Overall, CHIVE automatically discovers explanations of naturally-occurring LLM behaviors, enabling us to evaluate and improve methods for explaining LLM behaviors."], "abstract": "Many areas of AI research, such as language model interpretability and chain of thought faithfulness, seek to explain model behaviors. But what constitutes a \"good\" explanation? In this work, we evaluate explanations through the lens of counterfactual simulatability-whether the explanation is useful for predicting model behaviors on related counterfactual inputs. To this end, we introduce CHIVE (Counterfactual Hypothesis Investigation Via Edits), a novel agentic pipeline that identifies unexpected model behaviors in the wild and investigates them with counterfactual prompt edits. This yields thousands of high-quality explanations for naturally-occurring model behaviors along with supporting counterfactual evidence. We apply CHIVE in two ways. First, we evaluate whether common LLM interpretability techniques improve an agent's ability to predict counterfactual model behaviors. Surprisingly, we find no uplift from any of the interpretability techniques studied. Second, we use CHIVE to generate training data. We find that training models to predict outcomes of CHIVE-generated counterfactual experiments generalizes to various out-of-distribution settings. Overall, CHIVE automatically discovers explanations of naturally-occurring LLM behaviors, enabling us to evaluate and improve methods for explaining LLM behaviors.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 102180, "fulltext_methods": ["Chain-of-Thought", "Distillation", "Quantization", "Long Context", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "Tool Use", "Computer Use", "Deep Research"], "fulltext_benchmarks": ["MMLU", "DROP"], "fulltext_models": ["GPT-4o", "GPT-5.5", "Claude Opus 4.6", "Claude Fable 5 in an overnight goal loop that optimized prompts and hyperparameters for each tool", "Claude sonnet 4.5 system card. Note", "claude-sonnet-4-5-system-cardAccessed", "Claude mythos 5 system card. Note", "claude-fable-5-mythos-5-system-cardAccessed", "Claude Sonnet 4.6 as the auditor", "Gemini-3.1-Pro as the predictor agent in place of Opus 4.8 gives the same qualitative results"], "matched_methods": ["Chain-of-Thought", "Interpretability", "Multi-Agent"], "checked_on": "2026-08-19T03:31:07.456108+00:00"}}
{"id": "evg-33eb894710bfa6ef", "title": "TDD-Agent: Test-Driven Reasoning for Code Generation", "authors": ["Hongyue Yu", "Kefan Li", "Jiakun Li", "Hongzheng Chai", "Yuan Yuan", "Rui He", "Junyi Wei"], "year": 2026, "published": "2026-08-17T15:52:04Z", "updated": "2026-08-17T15:52:04Z", "arxiv_id": "http://arxiv.org/abs/2608.16742v1", "url": "https://arxiv.org/abs/2608.16742v1", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": ["LiveCodeBench"], "models": ["yielding"], "key_results": ["Large Language Models (LLMs) have achieved remarkable progress in code generation, yet ensuring correctness in complex, repository-level tasks remains challenging.", "We first isolate the effect of test-first reasoning through a prompt variant TDD-prompt on LiveCodeBench, where it consistently improves upon reasoning-based prompting baselines.", "Building on this finding, we evaluate the full TDD-Agent framework on RepoEval, a repository-level benchmark, and show that it consistently outperforms retrieval-based and agent-based baselines."], "abstract": "Large Language Models (LLMs) have achieved remarkable progress in code generation, yet ensuring correctness in complex, repository-level tasks remains challenging. Existing approaches often use generated tests as static post-hoc validators, which limits their ability to guide implementation and may introduce misleading feedback when the tests themselves are incomplete or incorrect. In this paper, we introduce TDD-Agent, which operationalizes the test-driven development paradigm for code generation. TDD-Agent first prompts the model to generate executable tests, encouraging it to clarify expected behaviors before implementation, and then performs iterative dual-track refinement over both the generated code and tests using execution feedback. We first isolate the effect of test-first reasoning through a prompt variant TDD-prompt on LiveCodeBench, where it consistently improves upon reasoning-based prompting baselines. Building on this finding, we evaluate the full TDD-Agent framework on RepoEval, a repository-level benchmark, and show that it consistently outperforms retrieval-based and agent-based baselines. Additional analyses show that iterative refinement improves not only code correctness but also the effectiveness of the generated tests, yielding higher pass rates, coverage, and mutation scores, suggesting that tests can serve as evolving reasoning artifacts rather than fixed validators. Our source code is available at https://anonymous.4open.science/r/TDD-Agent-Framework-6370/.", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 98895, "fulltext_methods": ["RLVR / GRPO", "Chain-of-Thought", "MoE", "Multi-Agent", "Memory / RAG", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["HumanEval", "MBPP", "SWE-bench", "LiveCodeBench", "COCO", "DROP"], "fulltext_models": ["GPT-4", "GPT-5-mini", "GPT-5", "gpt-5-miniOpenAI", "gpt-5-mini-2025-08-07", "GPT-5-mini.", "Claude", "claude 3 model family", "Gemini", "gemini research and technology. External Links"], "matched_methods": [], "checked_on": "2026-08-19T03:31:10.305806+00:00"}}
{"id": "evg-23136fc8736a1a89", "title": "An Analytical-Prior Framework for Data-Efficient Prediction of Sound-Reduction Frequencies in Rectangular Side-Branch Helmholtz Resonators", "authors": ["Jiaming Li"], "year": 2026, "published": "2026-08-17T17:53:14Z", "updated": "2026-08-17T17:53:14Z", "arxiv_id": "http://arxiv.org/abs/2608.16873v1", "url": "https://arxiv.org/abs/2608.16873v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["This study develops an analytical-prior learning framework that reuses a low-cost analytical model to improve data efficiency under limited high-fidelity simulation budgets.", "The analytical model achieved a mean absolute error (MAE) of 1.333 Hz.", "Direct support vector regression (SVR) achieved 3.375 Hz, while residual SVR reduced the MAE to 0.426 Hz."], "abstract": "High-fidelity finite-element simulations can provide accurate numerical predictions for side-branch resonators, but large simulation datasets are expensive to generate and purely data-driven surrogates may become unreliable when simulation-labelled data are scarce. This study develops an analytical-prior learning framework that reuses a low-cost analytical model to improve data efficiency under limited high-fidelity simulation budgets. Two complementary routes are considered. When the analytical model remains available at inference, it is retained as an explicit baseline and the simulation data are used to learn only the analytical-to-simulation discrepancy. When a self-contained predictor is required, the analytical mapping is first distilled from abundant low-cost evaluations into a learned prior and then calibrated with the limited simulation data. The framework is evaluated on rectangular side-branch Helmholtz resonators using 86 simulation-labelled geometries and 8,998 non-overlapping analytical-only geometries. The analytical model achieved a mean absolute error (MAE) of 1.333 Hz. Direct support vector regression (SVR) achieved 3.375 Hz, while residual SVR reduced the MAE to 0.426 Hz. A direct multilayer perceptron (MLP) achieved 1.109 Hz, whereas analytical-prior pretraining reduced the error to 0.556 Hz with frozen-prior residual adaptation and 0.371 Hz with full-model fine-tuning. Across training budgets of 20 to 70 simulation-labelled cases, both analytical correction and analytical-prior pretraining consistently improved data efficiency relative to direct learning. These results show that analytical prior information can substantially improve high-fidelity prediction when simulation data are scarce, with explicit correction and prior distillation serving complementary deployment needs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": false, "verification": {"source": "unavailable", "status": "unavailable", "checked_on": "2026-08-19T03:31:25.045082+00:00"}}
{"id": "evg-a5642de1ba7f8289", "title": "Every Coin Has Two Sides: On the Dual Nature of Generalization in On-Policy Distillation of Large Language Models", "authors": ["Zhaoyi Li", "Deyang Kong", "Yuan Wei", "Evan Yang", "Ranran Shen", "Mahardika Krisna Ihsani", "Ming Yang", "Wei Zhang", "Chuan Hao", "Jian Yang", "Ran Tao", "Bryan Dai", "Shikun Zhang", "Wei Ye", "Ying Wei", "Defu Lian"], "year": 2026, "published": "2026-08-17T14:46:53Z", "updated": "2026-08-17T14:46:53Z", "arxiv_id": "http://arxiv.org/abs/2608.16647v1", "url": "https://arxiv.org/abs/2608.16647v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["yields"], "key_results": ["On-policy distillation (OPD) transfers teacher capabilities by supervising trajectories sampled from the student's own policy, yet its generalization behavior remains poorly understood, as most studies evaluate OPD on a single domain and on benchmarks close to the training data."], "abstract": "On-policy distillation (OPD) transfers teacher capabilities by supervising trajectories sampled from the student's own policy, yet its generalization behavior remains poorly understood, as most studies evaluate OPD on a single domain and on benchmarks close to the training data. We present a controlled study that varies one generalization factor at a time, from in-domain distribution shifts to cross-domain transfer and the multi-teacher setting. We find that OPD transfers a teacher's reasoning behavior rather than its answers to particular problems: training difficulty barely matters, and even problems the teacher never solves are useful. Transfer depends strongly on the origin relationship between teacher and student: same-origin pairs bring the student close to the teacher across languages, reasoning horizons, and even other domains, whereas cross-origin pairs mostly fit the trained distribution. This broad reach is a double-edged sword: since routing prompts to domain experts cannot confine each teacher's influence, combining them yields a mixture-dependent seesaw among their capabilities. These results clarify when OPD generalizes and offer a useful perspective for diagnosing multi-teacher OPD.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 77112, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Chain-of-Thought", "Distillation", "Long Context", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["GSM8K", "MATH", "AIME", "GPQA", "LiveCodeBench", "IFEval"], "fulltext_models": ["o3", "o3-mini", "O3-mini-Level-1cf81902c14680b3bee5eb349a512a51Notion", "Llama-nemotron", "Llama-Nemotron-Post-Training-Dataset", "Qwen3-32B", "Qwen-1.5B", "Qwen-14B", "Qwen3-8B-SFT", "Qwen3-4B"], "matched_methods": ["Distillation"], "checked_on": "2026-08-19T03:31:29.639933+00:00"}}
{"id": "evg-3d07432eda2f93e1", "title": "Palmyra x6 Technical Report: An Agentic, Tool-Use Model Post-Trained via Anchored Supervised Fine-Tuning", "authors": ["Peng Du", "Kiran Kamble", "Rakshith Vasudev", "Zhizhuo Yang", "Rohith Nadimpally", "Arjun Krishna", "Waseem Alshikh", "Daniel M. Bikel"], "year": 2026, "published": "2026-08-17T14:21:03Z", "updated": "2026-08-17T14:21:03Z", "arxiv_id": "http://arxiv.org/abs/2608.16620v1", "url": "https://arxiv.org/abs/2608.16620v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "Safety / Jailbreak", "Multi-Agent", "Tool Use"], "benchmarks": [], "models": ["Palmyra x6 Technical Report", "Palmyra x6 is a large language model optimized for use with enterprise-oriented agentic tasks. The model was built by post-training a Mixture-of-Experts base model with Anchored Supervised Fine-Tuning on a compact corpus of verified"], "key_results": ["The model shows substantial gains over the previous default model for Writer Agent, and compares favorably with several recent models on public benchmarks, scoring the highest on BFCL Core at $0.785$ and posts the highest six-benchmark mean of the cohort."], "abstract": "Palmyra x6 is a large language model optimized for use with enterprise-oriented agentic tasks. The model was built by post-training a Mixture-of-Experts base model with Anchored Supervised Fine-Tuning on a compact corpus of verified, synthetic tool-use trajectories, optimized with a Muon + Adam hybrid. The recipe is deliberately conservative and deliberately controlled: 626 trajectories, a single epoch, a low learning rate, and a KL anchor to the frozen base. The model shows substantial gains over the previous default model for Writer Agent, and compares favorably with several recent models on public benchmarks, scoring the highest on BFCL Core at $0.785$ and posts the highest six-benchmark mean of the cohort. Furthermore, the model has shown itself to be competitive or leading relative to comparators in our bias and safety evaluations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 38007, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "MoE", "Distillation", "Quantization", "Long Context", "Preference Optimization", "Safety / Jailbreak", "Multi-Agent", "Tool Use", "Memory / RAG", "Quant / Trading"], "fulltext_benchmarks": ["TruthfulQA"], "fulltext_models": ["GPT-5.6-Luna", "DeepSeek", "DeepSeek-AI", "DeepSeek-V2", "DeepSeek-V3", "DeepSeek-V3.2", "V3", "V3.2", "phic", "GLM-5.2"], "matched_methods": ["MoE", "Multi-Agent", "Safety / Jailbreak", "Tool Use"], "checked_on": "2026-08-19T03:31:33.302840+00:00"}}
{"id": "evg-303ecb58b104f4af", "title": "Pallas: A Proactive KV Cache Migration Framework for LLM Inference in AI-RAN", "authors": ["Tianhang Ding", "Jianchun Liu", "Hongli Xu"], "year": 2026, "published": "2026-08-17T12:16:09Z", "updated": "2026-08-17T12:16:09Z", "arxiv_id": "http://arxiv.org/abs/2608.16477v1", "url": "https://arxiv.org/abs/2608.16477v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache"], "benchmarks": [], "models": [], "key_results": ["Across three LLMs and $100$--$500~\\mathrm{Mbps}$ inter-gNB links, our vLLM-based prototype reduces average SIT by factors of $2.28$--$89.68$ over target-side recovery approaches and lowers average ITL by $16.0\\%$--$50.0\\%$ compared with source-side forwarding."], "abstract": "AI-RAN brings large language model (LLM) serving close to mobile users, but cellular handover can separate an active request from its inference state: the user attaches to a target base station (gNB) while the large and growing key-value (KV) cache remains at the source. Retaining inference at the source preserves service continuity but persistently increases inter-token latency (ITL), whereas recovering the state at the target restores serving locality but requires KV-cache transfer, recomputation, or a combination of both only after handover, directly prolonging service interruption time (SIT).\n  This work presents Pallas, a \\textit{proactive} KV-cache migration framework that prepares the inference state at the predicted target before handover, in parallel with ongoing source-side inference and token delivery. At the preparation trigger, Pallas partitions the token sequence into a stable historical prefix and an evolving suffix. The target reconstructs the prefix through local prefill, while the source streams the KV blocks generated for the suffix. At handover, the target assembles both portions into an up-to-date KV cache and resumes decoding locally, leaving only unfinished preparation to contribute to SIT. An online scheduler selects the \\textit{prefetching window}, which determines how early preparation begins before handover, based on mobility predictions and runtime telemetry. Across three LLMs and $100$--$500~\\mathrm{Mbps}$ inter-gNB links, our vLLM-based prototype reduces average SIT by factors of $2.28$--$89.68$ over target-side recovery approaches and lowers average ITL by $16.0\\%$--$50.0\\%$ compared with source-side forwarding.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 66088, "fulltext_methods": ["RLVR / GRPO", "Quantization", "KV Cache", "Long Context", "Memory / RAG", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["Llama-3-8B", "Llama3.1-8B", "Qwen3-32B", "Qwen3-14B", "Qwen3-32B-Instruct-AWQ", "Qwen3", "ying"], "matched_methods": ["KV Cache"], "checked_on": "2026-08-19T03:31:36.458078+00:00"}}
{"id": "evg-07a66cab7aca15b7", "title": "Mint-Agent: Introducing Finance-Native Agentic Foundation Models", "authors": ["Mint-Agent Team", "B. Zhang", "Yaze Geng", "Lei Tang", "Yaoyang Yi", "Zonghan Wu", "Yifan Hu", "Kun Wang", "Qingsong Wen", "Yilei Shao"], "year": 2026, "published": "2026-08-17T10:38:54Z", "updated": "2026-08-17T10:38:54Z", "arxiv_id": "http://arxiv.org/abs/2608.16386v1", "url": "https://arxiv.org/abs/2608.16386v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "Distillation", "Preference Optimization", "Multi-Agent", "Quant / Trading"], "benchmarks": ["AgentBench"], "models": ["GPT-5.6-Sol", "Claude-Opus-4.8 by 3.66 and 3.00 points", "yields"], "key_results": ["Across professional financial benchmarks, our models demonstrate two defining strengths: (1) Reliability: Mint-Ag achieves 98.33% on RFC-Bench, surpassing GPT-5.6-Sol and Claude-Opus-4.8 by 3.66 and 3.00 points; and (2) Executability: Mint-Cu reaches 69.86% on FinSearchComp T2, outperforming Agents-A1-35B and Nex-N2-mini by 22.83 and 12.78 points, while Mint-Ag achieves 76.00% and 60.49% on FinanceAgentBench v1.1 and v2, respectively."], "abstract": "Financial agents must do more than recall domain knowledge: they must be both reliable, executing precise operations over grounded evidence, and executive, sustaining long-horizon research whose conclusions remain auditable. We present Mint-Agent, a family of finance-native agentic models designed around these two scales of financial intelligence. Mint-Agent is built upon three pillars: data, harness, and algorithm. Our data engine constructs clean, specialized tasks for atomic financial capabilities and long-horizon agentic execution from real-world financial sources. MintHarness enables stable interaction with open-ended environments and maintains auditable evidence trails across extended research trajectories. Our training recipe combines SFT, critical-step OPD, and RLVR to develop separate financial reasoning and agentic execution experts, which are then unified through model merging and multi-teacher on-policy distillation into compact, general-purpose financial agents. This pipeline yields two flagship models, Mint-Cu (9B) and Mint-Ag (27B). Across professional financial benchmarks, our models demonstrate two defining strengths: (1) Reliability: Mint-Ag achieves 98.33% on RFC-Bench, surpassing GPT-5.6-Sol and Claude-Opus-4.8 by 3.66 and 3.00 points; and (2) Executability: Mint-Cu reaches 69.86% on FinSearchComp T2, outperforming Agents-A1-35B and Nex-N2-mini by 22.83 and 12.78 points, while Mint-Ag achieves 76.00% and 60.49% on FinanceAgentBench v1.1 and v2, respectively. These results establish a path toward trustworthy financial intelligence in which domain expertise, long-horizon execution, and auditable evidence are jointly engineered as a unified foundation for frontier agentic models.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 84034, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Distillation", "Long Context", "Preference Optimization", "Multi-Agent", "Tool Use", "Memory / RAG", "Deep Research", "World Model"], "fulltext_benchmarks": ["AgentBench", "FinQA", "TAT-QA"], "fulltext_models": ["GPT-5.6-Sol", "GPT-5.6", "gpt-5-6", "Claude Opus 4.8 by 3.66 and 3.00 points", "Claude Opus 4.8", "Claude Opus 4.8 49.71 90.00 95.33 66.00 45.66 51.16 GPT-5.6-Sol 51.57 86.67 94.67 66.00 41.10 50.58 GLM-5.2 47.57 88.00 94.33 58.00 68.49 34.88 MiniMax-M3 40.71 86.00 91.33 70.00 60.27 30.23 Kimi-K2.7-Code 46.57 82.00 92.67 62.00 56.62 25.00 DeepSeek-V4-Pro 44.00 87.33 91.33 60.00 60.73 31.40 DeepSeek-V4-Flash 45.71 88.67 89.67 72.00 59.82 25.00 MiMo-V2.5-Pro 45.00 87.33 91.33 66.00 58.45 26.16 Qwen3.7-Plus 48.43 88.00 93.33 64.00 62.10 31.98 Open-source models Agents-A1-35B 41.86 84.00 90.00 42.00 47.03 11.63 Nex-N2-mini 27.43 69.33 83.33 38.00 57.08 20.35 ASearcher-32B 28.14 81.33 84.00 24.00 42.92 4.07 OpenThinkerAgent-32B 25.00 66.67 71.00 34.00 36.07 14.53 OpenResearcher-30B-A3B 10.71 50.00 22.00 42.00 53.88 16.86 Tongyi-DeepResearch-30B-A3B 19.57 42.67 57.67 58.00 72.15 18.02 Agent systems Codex", "Claude opus 4.8. Official model page", "claude", "Gemini-3.5-Flash 50.71 85.33 92.33 62.00 36.53 44.19 Claude Opus 4.8 49.71 90.00 95.33 66.00 45.66 51.16 GPT-5.6-Sol 51.57 86.67 94.67 66.00 41.10 50.58 GLM-5.2 47.57 88.00 94.33 58.00 68.49 34.88 MiniMax-M3 40.71 86.00 91.33 70.00 60.27 30.23 Kimi-K2.7-Code 46.57 82.00 92.67 62.00 56.62 25.00 DeepSeek-V4-Pro 44.00 87.33 91.33 60.00 60.73 31.40 DeepSeek-V4-Flash 45.71 88.67 89.67 72.00 59.82 25.00 MiMo-V2.5-Pro 45.00 87.33 91.33 66.00 58.45 26.16 Qwen3.7-Plus 48.43 88.00 93.33 64.00 62.10 31.98 Open-source models Agents-A1-35B 41.86 84.00 90.00 42.00 47.03 11.63 Nex-N2-mini 27.43 69.33 83.33 38.00 57.08 20.35 ASearcher-32B 28.14 81.33 84.00 24.00 42.92 4.07 OpenThinkerAgent-32B 25.00 66.67 71.00 34.00 36.07 14.53 OpenResearcher-30B-A3B 10.71 50.00 22.00 42.00 53.88 16.86 Tongyi-DeepResearch-30B-A3B 19.57 42.67 57.67 58.00 72.15 18.02 Agent systems Codex", "Gemini-3.5-Flash"], "matched_methods": ["Distillation", "Multi-Agent", "Preference Optimization", "Quant / Trading", "RLVR / GRPO"], "checked_on": "2026-08-19T03:31:39.494746+00:00"}}
{"id": "evg-e3afd29b7519de95", "title": "Step-Level On-Policy Distillation: Interpolating Between On-Policy Distillation and Supervised Fine-Tuning", "authors": ["Changhui Sun", "Lanbo Liu", "Hang Lei", "Tong Ling", "Jiahang Xie", "Zhiyong Zheng", "Yujia Wang", "Hao Liu", "Feng Xiao", "Lu Liu", "Yanlong Du", "Zifeng Cheng", "Ziwei Jiang", "Qing Gu"], "year": 2026, "published": "2026-08-17T09:37:39Z", "updated": "2026-08-17T09:37:39Z", "arxiv_id": "http://arxiv.org/abs/2608.16333v1", "url": "https://arxiv.org/abs/2608.16333v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["This approach has achieved strong empirical gains and can often surpass conventional off-policy distillation with substantially less data.", "We show that, at different limits of step length, SOPD reduces to SFT or approximates OPD.", "Across both reasoning and agent tasks, SOPD substantially outperforms conventional SFT and OPD."], "abstract": "On-policy distillation (OPD) aligns a student model with a teacher's logit distribution on student-generated trajectories. This approach has achieved strong empirical gains and can often surpass conventional off-policy distillation with substantially less data. However, standard token-level OPD can provide only fragmented corrections along an erroneous student trajectory and cannot unfold a complete and correct repair path. Motivated by this limitation, we propose \\emph{Step-Level On-Policy Distillation} (SOPD), which combines the long-horizon correction of supervised fine-tuning (SFT) with the on-policy advantage of OPD to provide step-level supervision over complete student-generated trajectories. We show that, at different limits of step length, SOPD reduces to SFT or approximates OPD. Compared with SFT, the teacher responses in SOPD are conditioned on student trajectories and therefore align more closely with student-visited states; compared with OPD, SOPD provides longer-horizon corrections rather than fragmented token-level guidance. Across both reasoning and agent tasks, SOPD substantially outperforms conventional SFT and OPD. For example, on ALFWorld, SOPD improves the average success rate by 13.4 points over Vanilla OPD. We hope this work offers a new perspective for future research on distillation methods.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 37271, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Distillation", "Preference Optimization", "Safety / Jailbreak"], "fulltext_benchmarks": ["MATH", "AIME"], "fulltext_models": ["Qwen2.5-7B", "Qwen2.5-3B-Instruct", "Qwen2.5-7B-RL", "Qwen2.5-3B", "Qwen3-4B", "Qwen3-4B-Non-Thinking-RL-Math.", "Qwen3-4B-Non-Thinking-RL-Math", "Qwen3", "Qwen3-4B-Non-Thinking-RL-Math-Step500", "Qwen2.5-7B-Instruct-ALFWorld"], "matched_methods": ["Distillation", "Preference Optimization"], "checked_on": "2026-08-19T03:31:45.858558+00:00"}}
{"id": "evg-a4f9540813ababf3", "title": "Deep Thought Alignment: Trajectory-Level Latent Distillation for Video Reasoning", "authors": ["Ao Shen", "Yongheng Zhang", "Yinghui Li", "Manning Wang", "Di Yin", "Xing Sun"], "year": 2026, "published": "2026-08-17T09:23:49Z", "updated": "2026-08-17T09:23:49Z", "arxiv_id": "http://arxiv.org/abs/2608.16316v1", "url": "https://arxiv.org/abs/2608.16316v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Search / MCTS", "Distillation", "Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["Experiments on six video reasoning benchmarks show that Latent-OPD consistently outperforms output-only OPD.", "Notably, the improvements are particularly pronounced in scenarios with limited frames, long videos, or tasks requiring complex evidence aggregation."], "abstract": "Large Multimodal Models (LMMs) for video reasoning have long been hindered by the high computational cost of processing vast amounts of visual information. This dilemma motivates the transfer of the reasoning capabilities of large models to smaller, more efficient ones. On-Policy Distillation (OPD) offers a promising solution by matching output-token distributions along student-generated trajectories. However, video reasoning often depends on evidence accumulated across multiple frames. In this context, output-level supervision only captures information expressed through token predictions and does not directly constrain the latent representations formed during reasoning. To address this limitation, we propose Latent-OPD, which augments OPD with trajectory-level latent distillation. Specifically, our method focuses on the position at the end of each trajectory, where hidden states effectively summarize the accumulated visual evidence and reasoning context. Furthermore, we introduce a progressive teacher-lookahead strategy, which aligns middle-to-late student layers with increasingly deeper teacher layers. Experiments on six video reasoning benchmarks show that Latent-OPD consistently outperforms output-only OPD. Notably, the improvements are particularly pronounced in scenarios with limited frames, long videos, or tasks requiring complex evidence aggregation. These results establish Latent-OPD as a highly effective approach to frame-efficient video reasoning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 58576, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Chain-of-Thought", "Distillation", "Long Context", "Preference Optimization", "Multi-Agent", "World Model", "VLM"], "fulltext_benchmarks": ["MMMU", "Video-MME"], "fulltext_models": ["llama 2", "Llama-vid", "Qwen3.5-9B-CoT", "Qwen3.5-9B-SFT", "Qwen3.5-9B", "Qwen3.5-9B-Base", "Qwen3.5-27B-Base", "Qwen3.5-4B", "Qwen3.5-4B-Base", "Qwen3-vl"], "matched_methods": ["Distillation", "Preference Optimization", "Search / MCTS", "VLM"], "checked_on": "2026-08-19T03:31:49.338919+00:00"}}
{"id": "evg-078552f76f82ac6c", "title": "EMS Coreset: An Efficient Expectation-Maximization Algorithm for Sinkhorn Coreset", "authors": ["Haoyun Yin", "Chuanhui Liu", "Xiao Wang"], "year": 2026, "published": "2026-08-17T04:46:34Z", "updated": "2026-08-17T04:46:34Z", "arxiv_id": "http://arxiv.org/abs/2608.16101v1", "url": "https://arxiv.org/abs/2608.16101v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["Across synthetic and real-world benchmarks, the proposed method achieves competitive or improved approximation quality while substantially reducing runtime compared to Wasserstein- and standard Sinkhorn-based coreset selection, especially at large scale."], "abstract": "Coresets distill large datasets into small, representative subsets for efficient downstream learning. Yet Optimal Transport (OT)-based selection typically requires intensive computation of transport plans, limiting scalability. We introduce a scalable Sinkhorn coreset method that permits closed-form updates of the entropically regularized OT coupling by allowing non-uniform coreset weights. This produces centroids that generalize k-means via soft assignments. We establish asymptotic consistency of the selected measure and Lipschitz stability to data perturbations, providing accuracy and robustness guarantees. Across synthetic and real-world benchmarks, the proposed method achieves competitive or improved approximation quality while substantially reducing runtime compared to Wasserstein- and standard Sinkhorn-based coreset selection, especially at large scale.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 60678, "fulltext_methods": ["Distillation", "Quantization", "Interpretability", "Memory / RAG", "Deep Research", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["Yin", "yin164", "ying", "yielding", "yields", "yield"], "matched_methods": ["Distillation"], "checked_on": "2026-08-19T03:31:52.743975+00:00"}}
{"id": "evg-96f1adec4d92c7f3", "title": "Eigenanalysis framework for autoregressive neural emulators of multi-scale chaotic dynamics", "authors": ["Conrad Ainslie", "Pedram Hassanzadeh", "Michael W. Mahoney", "Ashesh Chattopadhyay"], "year": 2026, "published": "2026-08-17T04:30:03Z", "updated": "2026-08-17T04:30:03Z", "arxiv_id": "http://arxiv.org/abs/2608.16084v1", "url": "https://arxiv.org/abs/2608.16084v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG", "nlin.CD", "physics.comp-ph"], "pillar": "Efficient Training & Inference", "methods": ["Test-time Scaling", "Quant / Trading"], "benchmarks": [], "models": ["yielding"], "key_results": ["Neural autoregressive models have rapidly emerged as powerful emulators of high-dimensional chaotic systems, yet their long-term instability and error growth remain poorly understood, leading to ad-hoc solutions."], "abstract": "Neural autoregressive models have rapidly emerged as powerful emulators of high-dimensional chaotic systems, yet their long-term instability and error growth remain poorly understood, leading to ad-hoc solutions. Here, we develop an eigenanalysis framework that reveals the dynamical origin of this error growth. By analyzing the Jacobian of the learned one-step update map with respect to the state, we show how inference-time error growth, and thus model stability, is governed by its spectral radius. Direct-step architectures (models that predict the next state from the previous one) generically admit unstable eigenvalues with magnitudes exceeding one, explaining the rapid divergence of these widely used models. In contrast, integration-constrained models (where the time derivative is estimated and integrated with a higher-order integrator) collapse their eigenspectrum onto the unit circle, yielding neutral stability and a universal linear error-scaling law. The largest eigenvalue of this Jacobian provides an architecture-agnostic, a priori diagnostic of short-term skill, long-term stability, and spectral bias, without requiring an expensive rollout. Leveraging this theory, we introduce a stability-promoting loss that explicitly regularizes Jacobian-driven error amplification, improving both forecast accuracy and dynamical robustness. Demonstrated across $29$ models spanning two architectures, several explicit and implicit integrators, and multiple loss functions on the Kuramoto-Sivashinsky system, our results establish a theoretical foundation for the design and evaluation of neural emulators of chaotic multi-scale dynamics. More broadly, our framework is a step toward the kind of a priori stability analysis that numerical analysis provides for discretizations of differential equations and that scientific machine learning currently lacks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 61385, "fulltext_methods": ["Test-time Scaling", "Interpretability", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["yielding", "ying", "yield", "yields"], "matched_methods": ["Quant / Trading", "Test-time Scaling"], "checked_on": "2026-08-19T03:31:57.255415+00:00"}}
{"id": "evg-344abe3337c1b5ca", "title": "GOD: Enhancing Generalization via Deep Grafting for Sequential Recommendation", "authors": ["WooJoo Kim", "JunYoung Kim", "JaeHyung Lim", "HwanJo Yu"], "year": 2026, "published": "2026-08-17T04:11:49Z", "updated": "2026-08-17T04:11:49Z", "arxiv_id": "http://arxiv.org/abs/2608.16073v1", "url": "https://arxiv.org/abs/2608.16073v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["In this paper, we propose Graft-Oriented Distillation (GOD), a component-level distillation framework for improved generalization through grafting.", "Across three real-world datasets, GOD outperforms state-of-the-art baselines by up to 13.92%."], "abstract": "Sequential recommenders often struggle with sparse and noisy histories, limiting generalization to unseen interactions. Knowledge distillation mitigates this by transferring dense supervision from a teacher to a student. However, most distillation methods run teacher and student independently, then match student outputs or representations to the teacher. Such supervision entangles student-component effects, blurring whether weak generalization stems from unreliable embeddings, overfitted encoding, or co-adaptation to sparse histories. In this paper, we propose Graft-Oriented Distillation (GOD), a component-level distillation framework for improved generalization through grafting. Grafting denotes replacing selected frozen-teacher components with trainable student counterparts to build hybrid source models. GOD uses these hybrid models to evaluate student embeddings with the teacher encoder and the student encoder with teacher embeddings, providing component-level feedback. At inference, GOD uses only the student, incurring no additional cost. Across three real-world datasets, GOD outperforms state-of-the-art baselines by up to 13.92%.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 35441, "fulltext_methods": ["Distillation", "Preference Optimization", "Memory / RAG"], "fulltext_benchmarks": [], "fulltext_models": ["r11", "yield", "yielding"], "matched_methods": ["Distillation"], "checked_on": "2026-08-19T03:32:04.392483+00:00"}}
{"id": "evg-49b1e9c4e2f58bf5", "title": "A Scalable Pipeline for LLM-Teacher Distillation Labeling: Work-Stealing Job Scheduling and Memory-Aware GPU Concurrency", "authors": ["Ravi Satya Durga Prasad Yenugula"], "year": 2026, "published": "2026-08-17T00:02:25Z", "updated": "2026-08-17T00:02:25Z", "arxiv_id": "http://arxiv.org/abs/2608.15975v1", "url": "https://arxiv.org/abs/2608.15975v1", "primary_category": "cs.DC", "categories": ["cs.DC", "cs.AI", "cs.CL", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Memory / RAG"], "benchmarks": [], "models": ["yields"], "key_results": ["Third, a relabeling benchmark methodology in which the teacher relabels a public dataset that already has gold labels, so quality reduces to an agreement measurement and cost follows from measured throughput."], "abstract": "Labeling large text corpora with LLM teachers has become a practical route to training data at scale. At millions of items, hand-labeling every batch is not feasible, and two questions dominate: what label quality a teacher buys per dollar, and how to keep a fleet of GPU workers busy under skewed, failure-prone workloads. We present a simple, reproducible pipeline that addresses both. First, a work-stealing ring pool: each worker owns a queue, drains it first, and then steals from ring successors, with exactly-once task claims via atomic conditional writes and crash tolerance via stale-claim sweeping. The claim protocol requires only a compare-and-set primitive from its storage layer; we implement it on a single SQLite file, which makes the reference implementation dependency-free and the experiments reproducible on one machine. Second, a memory-aware concurrency rule that sizes per-node parallelism by how many model copies fit on the GPU, so the same code runs safely across device sizes. Third, a relabeling benchmark methodology in which the teacher relabels a public dataset that already has gold labels, so quality reduces to an agreement measurement and cost follows from measured throughput. Under skewed load the pool sustains up to 3.4 times the throughput of static sharding while matching it at zero skew, loses 0 of 2,000 tasks when half the workers are killed mid-run (static sharding loses 953), and yields measured quality and cost points for an instruction-tuned teacher on irony and sentiment tasks. All experiments run on public data and commodity hardware; code, tests, and run logs are released.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00"}
{"id": "evg-83f07bb7124dab50", "title": "SEER: Long-Context Reasoning via Selective Visual-Text Compression", "authors": ["Jiawei Xu", "Zhilin Zhai", "Jinrui Fang", "Ruohan Xu", "Mingfei Lu", "Yi Zhang", "Guanchu Wang", "Tianlong Chen", "Ying Ding"], "year": 2026, "published": "2026-08-16T23:30:41Z", "updated": "2026-08-16T23:30:41Z", "arxiv_id": "http://arxiv.org/abs/2608.15962v1", "url": "https://arxiv.org/abs/2608.15962v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["Long Context", "VLM"], "benchmarks": [], "models": ["Qwen3-8B"], "key_results": ["Experiments on long-context benchmarks show that SEER improves extraction precision through selective text retrieval while retaining average prompt-token savings relative to full-text baselines.", "On LongBench, SEER achieves 51.11% average accuracy, outperforming the visual-text baseline Glyph-9B by 2.33 points and Qwen3-8B by 3.49 points."], "abstract": "Long-context reasoning remains computationally expensive for large language models due to the quadratic complexity of attention over text tokens. Visual-text compression offers a promising alternative by rendering text into images and processing them with vision-language models, often reducing token usage. However, existing approaches apply uniform compression regardless of query relevance, potentially sacrificing precision where detailed extraction is required. We present SEER, a framework that learns to select query-relevant images through visual scanning and retrieve textual content only where needed, combining the efficiency of visual compression with the precision of text-based reasoning. Through supervised fine-tuning on tool-interaction trajectories, SEER learns adaptive tool invocation for selection and retrieval. Experiments on long-context benchmarks show that SEER improves extraction precision through selective text retrieval while retaining average prompt-token savings relative to full-text baselines. On LongBench, SEER achieves 51.11% average accuracy, outperforming the visual-text baseline Glyph-9B by 2.33 points and Qwen3-8B by 3.49 points. Code can be accessed at https://github.com/jiaweixu98/SEER", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00"}
{"id": "evg-4fc7273824840e92", "title": "Data-Efficient and Interpretable Classification of Circulating Tumor Cell Phenotypes in Microfluidic Devices via Deep Learning", "authors": ["Serena Su", "Yifan Wang", "Senwei Liang"], "year": 2026, "published": "2026-08-17T17:51:49Z", "updated": "2026-08-17T17:51:49Z", "arxiv_id": "http://arxiv.org/abs/2608.16870v1", "url": "https://arxiv.org/abs/2608.16870v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["Experimental results demonstrate that SubSeq improves classification accuracy over the evaluated baseline and augmentation methods."], "abstract": "Accurate classification of circulating tumor cell (CTC) phenotypes can provide valuable information for assessing metastatic potential. Label free microfluidic devices provide a hydrodynamic obstacle course that transforms subtle biophysical characteristics of CTCs, including size and deformability, into distinct kinematic trajectories. However, the highly nonlinear fluid structure interactions governing these trajectories make the inverse problem of inferring cellular phenotype from trajectory data analytically intractable. While deep neural networks (DNNs) have emerged as a powerful approach for addressing this inverse problem, their effectiveness is constrained by the limited availability of trajectory data and the lack of physical interpretability.\n  To address these challenges, we propose an interpretable and data efficient DNN framework for trajectory based CTC classification. To mitigate the scarcity of data, we develop Subsequence (SubSeq), a targeted augmentation strategy that randomly extracts informative local trajectory segments during training to promote learning from localized patterns. We further apply Gradient Weighted Class Activation Mapping to identify the trajectory features and physical regions of the microfluidic device that drive model predictions. Experimental results demonstrate that SubSeq improves classification accuracy over the evaluated baseline and augmentation methods. Furthermore, interpretability analysis suggests that localized trajectory segments contain substantial biophysical information relevant to accurate classification. This provides justification for SubSeq and also highlights the redundancy of full-length trajectories. More broadly, the proposed framework views microfluidic geometries as physical encoders of cellular mechanical properties, providing mechanistic insights that may inform the future design of diagnostic devices.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 50467, "fulltext_methods": ["Interpretability", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["R1", "phin", "Yifan", "yifan.wang", "ying", "yields", "yielding"], "matched_methods": ["Interpretability"], "checked_on": "2026-08-19T03:32:06.748870+00:00"}}
{"id": "evg-353a4635eade0d7d", "title": "CaliBench: Are the Stochastic Dynamics of Video World Models Physically Calibrated?", "authors": ["Jonathan Sadeghi", "Jenny Seidenschwarz", "Jesse Allardice", "Sirish Srinivasan", "Benjamin Graham", "Jeffrey Hawke"], "year": 2026, "published": "2026-08-17T17:14:50Z", "updated": "2026-08-17T17:14:50Z", "arxiv_id": "http://arxiv.org/abs/2608.16829v1", "url": "https://arxiv.org/abs/2608.16829v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "World Model"], "benchmarks": [], "models": ["yielding"], "key_results": ["We release the protocol and a metric (mean normalised total variation, mnTV) for comparing new models against our results."], "abstract": "Video world models approximate the stochastic distribution of physical outcomes through generative sampling, but existing benchmarks score individual generations or compare distributions coarsely over a whole dataset, leaving the fine-grained aleatoric uncertainty of specific phenomena untested. We introduce CaliBench, which scores outcomes in a physically interpretable discrete space - a bin index, a die face, a suit, a colour - rather than a learned feature space such as in FID, so the distance from a known reference distribution is measured directly. We curate outcome spaces whose reference is known in closed form (binomial Galton boards, Bernoulli forks, uniform dice/cards/lottery, a skewed European-roulette colour), enabling an exact calibration test. We decompose performance into two orthogonal axes that a single accuracy metric conflates: scorability, the fraction of generations yielding a scoreable outcome, and calibration, the total variation distance from the reference on that sample. A chi-squared test assesses significance; as calibration is its null hypothesis it can evidence only miscalibration, and at N=32 per cell detects only large deviations. We apply it to nine scenes and six image-to-video models (WAN-2.7, SeeDance-2.0, HappyHorse-1.0, Veo 3.1, Runway Gen-4.5, Cosmos3-Super), 32 generations each. Models consistently concentrate probability mass on a few outcomes rather than reproducing the reference. Most scene-model combinations are significantly miscalibrated, in the extreme collapsing to one outcome, as Veo 3.1 does on dice. On roulette, generations often leave the ball ambiguously placed, giving several models low scorability. Performance varies by scene: no model dominates all nine. We release the protocol and a metric (mean normalised total variation, mnTV) for comparing new models against our results.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 92505, "fulltext_methods": ["Test-time Scaling", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Computer Use", "Deep Research", "World Model", "Video Generation", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["Gemini 3.1 Pro", "Gemini 3.1 Pro. Note", "gemini-models", "gemini-3-1-pro", "Gemini 3.1 Pro alongside the scene-specific prompt below", "phic", "phing", "Philosophical", "ying", "yielding"], "matched_methods": ["Interpretability", "World Model"], "checked_on": "2026-08-19T03:32:10.860863+00:00"}}
{"id": "evg-c3648fcc6d45ee81", "title": "Beyond $L_2$: Generalizing Abductive Latent Explanations to Diverse Prototype-Based Architectures", "authors": ["Jules Soria", "Alban Grastien", "Romain Xu-Darme", "Julien Girard-Satabin", "Zakaria Chihani", "Daniela Cancila"], "year": 2026, "published": "2026-08-17T16:20:20Z", "updated": "2026-08-17T16:20:20Z", "arxiv_id": "http://arxiv.org/abs/2608.16773v1", "url": "https://arxiv.org/abs/2608.16773v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Safety / Jailbreak"], "benchmarks": [], "models": ["ying"], "key_results": ["This leaves a critical gap: modern state-of-the-art architectures increasingly rely on non-Euclidean representations - such as spherical metrics, Gaussian densities, and dimensional projections - rendering current formal explanation methods incompatible."], "abstract": "Prototype-based neural networks are hailed as interpretable-by-design architectures. Recently, Abductive Latent Explanations (ALE) were introduced to provide formal, mathematically guaranteed explanations that leverage the intrinsic structure of these networks to ensure both predictive safety and human readability. ALEs rely on computing tight bounds on latent space distances to produce formal explanations. However, existing ALE formulations are rigidly confined to Euclidean latent spaces. This leaves a critical gap: modern state-of-the-art architectures increasingly rely on non-Euclidean representations - such as spherical metrics, Gaussian densities, and dimensional projections - rendering current formal explanation methods incompatible. In this work, we generalize the ALE framework to support non-Euclidean prototype architectures. For each geometric variant, we systematically derive how to either map the architecture to existing bounds or construct novel, architecture-specific bounding algorithms. We validate our theoretical constructions by computing subset-minimal formal explanations on fully trained image classifiers. By unifying these diverse models under a single formal framework, we enable the first rigorous, cross-architecture comparison of their interpretability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 44784, "fulltext_methods": ["Interpretability", "Safety / Jailbreak", "Deep Research", "Quant / Trading"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["phics", "ying", "yields", "yield", "yIA", "Palma", "bert", "t5"], "matched_methods": ["Interpretability", "Safety / Jailbreak"], "checked_on": "2026-08-19T03:32:17.406887+00:00"}}
{"id": "evg-221754ff9554a89c", "title": "A Deployment-Oriented and Resource-Efficient Neuro-Symbolic Framework for Explainable DDoS Detection in Operational Technology Networks", "authors": ["Mikiyas Alemayehu", "Mohamed Chahine Ghanem", "Hamza Kheddar", "Aohan Li", "J. J. Garcia-Luna-Aceves"], "year": 2026, "published": "2026-08-17T16:18:15Z", "updated": "2026-08-17T16:18:15Z", "arxiv_id": "http://arxiv.org/abs/2608.16769v1", "url": "https://arxiv.org/abs/2608.16769v1", "primary_category": "cs.CR", "categories": ["cs.CR"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Operational technology (OT) environments, including programmable logic controllers (PLCs), industrial control systems (ICS), and supervisory control and data acquisition (SCADA) systems, are increasingly targeted by distributed denial-of-service (DDoS) attacks."], "abstract": "Operational technology (OT) environments, including programmable logic controllers (PLCs), industrial control systems (ICS), and supervisory control and data acquisition (SCADA) systems, are increasingly targeted by distributed denial-of-service (DDoS) attacks. This paper presents a neuro-symbolic framework specifically designed for robust DDoS detection in these resource-constrained environments. The framework fuses a gated recurrent unit (GRU) neural network with a shallow decision tree as a symbolic component. The symbolic component alone provides a compact, interpretable rule set, while the fusion combines the strengths of both paradigms. The hybrid model is evaluated on three real-world benchmark DDoS datasets: CIC-DDoS2019, Edge-IIoTset, and CICIoT23. A unified comprehensive preprocessing pipeline including label mapping, numerical feature selection, robust scaling, and class balancing is applied. The fusion weight alpha and decision threshold are jointly optimised on validation data to maximise F1-score. The hybrid model attains 99.04% accuracy (MCC 0.97) on CIC-DDoS2019 and 98.61% accuracy (MCC 0.76) on CICIoT23, in both cases reducing the FNR below that of the pure-neural and pure-symbolic baselines; on the linearly separable Edge-IIoTset the shallow decision tree alone already reaches 100%, so this benchmark validates the preprocessing pipeline rather than the fusion. The principal gain of the fusion is a lower FNR at a controlled false-positive cost, which matters in operational technology, where a missed attack is more damaging than a false alarm. Model-only inference latency is sub-millisecond (0.58-0.79 milliseconds per sample) on a standard central processing unit; including on-device flow-feature extraction, the end-to-end path remains within a single-digit-millisecond budget, which is compatible with OT control-loop timing.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 80870, "fulltext_methods": ["RLVR / GRPO", "Distillation", "Quantization", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "Memory / RAG", "Deep Research"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["o17050412.", "Phil", "yields", "ying", "yielding", "yield", "bert", "T5"], "matched_methods": ["Interpretability"], "checked_on": "2026-08-19T03:32:21.480295+00:00"}}
{"id": "evg-13ec214ea33f31a7", "title": "The Ethical Decision Head: Operationalizing Normative Ethics in Autonomous Vehicles via Reinforcement Learning from Human Feedback", "authors": ["Thomas Mbrice", "Ammar Ali", "Sami Mian", "Khai Hern Low", "Eric Chen", "Arshia Aghajani", "Wolf Schäfer", "Amin Shirangi"], "year": 2026, "published": "2026-08-17T15:26:09Z", "updated": "2026-08-17T15:26:09Z", "arxiv_id": "http://arxiv.org/abs/2608.16710v1", "url": "https://arxiv.org/abs/2608.16710v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": ["philosophers"], "key_results": ["The EDH is trained via Proximal Policy Op- timization (PPO) [Schulman et al., 2017] against a Bradley- Terry reward model [Bradley and Terry, 1952] learned from pairwise human preference annotations over 200 collision- imminent scenarios.", "The Kantian condition, which reduces to a con- stant prediction task under the codebook, serves as a pipeline control: it confirms training stability and rules out infrastruc- ture failure as an explanation for the utilitarian result."], "abstract": "As autonomous vehicles (AVs) approach Level 4 and Level 5 operational capability [SAE International, 2018], their on- board decision systems must handle not only safety-critical locomotion but also their subsequent moral weight. This paper details the Ethical Decision Head (EDH), a deep re- inforcement learning (RL) framework that encodes ethical reasoning as a differentiable reward signal, enabling a pol- icy gradient agent to learn morally-aligned driving behavior in scenarios whose state representation is aligned with the CARLA simulation environment [Dosovitskiy et al., 2017]. Two normative frameworks are instantiated and evaluated: a Utilitarian framework minimizing total casualties and a Kan- tian framework enforcing course maintenance as a categori- cal imperative. The EDH is trained via Proximal Policy Op- timization (PPO) [Schulman et al., 2017] against a Bradley- Terry reward model [Bradley and Terry, 1952] learned from pairwise human preference annotations over 200 collision- imminent scenarios. Results reveal an asymmetry in the learnability of normative ethical frameworks under human su- pervision. The Kantian condition, which reduces to a con- stant prediction task under the codebook, serves as a pipeline control: it confirms training stability and rules out infrastruc- ture failure as an explanation for the utilitarian result. The Utilitarian agent learned something more unsettling: human raters rewarded self-sacrifice over casualty minimization, and the model learned that preference faithfully. This divergence between what humans prescribe in theory and what they re- ward in practice suggests that RLHF does not learn ethics as philosophers define it, but as humans live it.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 34095, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Safety / Jailbreak", "Computer Use", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["philosophers", "Philippa", "philosophy", "philosophically", "philosophical", "phic", "phically", "ying"], "matched_methods": ["Preference Optimization", "RLVR / GRPO", "Safety / Jailbreak"], "checked_on": "2026-08-19T03:32:24.354990+00:00"}}
{"id": "evg-7f5cacfffd223609", "title": "Semantic Bandits: In-Context Exploration-Exploitation is Biased by Semantic Priors", "authors": ["David Eric Austin", "Kaheer Suleman", "Jackie Chi Kit Cheung"], "year": 2026, "published": "2026-08-17T15:25:18Z", "updated": "2026-08-17T15:25:18Z", "arxiv_id": "http://arxiv.org/abs/2608.16707v1", "url": "https://arxiv.org/abs/2608.16707v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["phisticated"], "key_results": ["Large language models (LLMs) are increasingly deployed as decision-making agents in settings that require sophisticated environmental exploration."], "abstract": "Large language models (LLMs) are increasingly deployed as decision-making agents in settings that require sophisticated environmental exploration. However, existing work has raised questions about how LLMs actually balance exploration and exploitation. Unlike classical agents, LLM agents engage with tasks through natural language, exposing them to semantic information with no formal counterpart in the task structure. We introduce the semantic bandit, an extension of the multi-armed bandit setting that explicitly considers the textual labels assigned to actions, and use it to study how semantic priors --- inductive biases arising from associations between language and expected reward learned during pre-training, shape LLM exploration behaviour. We find that semantically informative action labels reduce exploration in favour of exploitation, improving performance when aligned with the reward structure and severely degrading it when misaligned. We further find that negative rewards trigger substantially more exploration than equivalent positive rewards, consistent with an expected-scale bias induced by reward conventions common in pre-training data. Overall, we argue that the use of language to define the environment and rewards introduces unavoidable biases derived from the fact that the model is trained on word co-occurence, with implications for the reliability and robustness of LLM agents in real-world decision-making settings.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 82472, "fulltext_methods": ["RLVR / GRPO", "Chain-of-Thought", "Distillation", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "Quant / Trading"], "fulltext_benchmarks": ["MATH", "WebArena"], "fulltext_models": ["GPT-4.", "Claude Code were used in the generation of plotting code and some minor edits to core experiment functionality. Appendix B Additional Experimental Results from Main Experiments This section contains more detailed results for the main experiments. This includes results using the", "Gemini 3.1 Flash Lite. Qwen3 and Gemini offer native support for thinking mode", "Gemini 3.1 Flash Lite Ordinal 0.13 0.27 0.15 0.17 0.15 0.22 World", "Gemini tended to explore quickly", "Gemini model and it may increasingly rely on semantic priors as task complexity increases. We provide a sample reasoning trace in Appendix B.5 which demonstrates that while semantic priors may be reduced in large frontier reasoning models", "Gemini. The impact of reward magnitude is unclear and highly non-monotonic. There was weak evidence for other threshold effects around 1.0", "Gemini-Clothing. Figure 5", "Gemini", "Gemini credits as well as the Digital Research Alliance of Canada and Mila for providing additional compute. References Auer et al."], "matched_methods": [], "checked_on": "2026-08-19T03:32:27.918595+00:00"}}
{"id": "evg-b4039c8c8db5da9e", "title": "Hide&Seek: Learning to Explain in an End-to-End Differentiable Network", "authors": ["Tal Ellinson", "Hadi Mohasel Afshar", "Sally Cripps"], "year": 2026, "published": "2026-08-17T15:10:05Z", "updated": "2026-08-17T15:10:05Z", "arxiv_id": "http://arxiv.org/abs/2608.16689v1", "url": "https://arxiv.org/abs/2608.16689v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Hide&Seek outperforms existing state-of-the-art models across a range of experiments and is fast to train.", "We achieve this by reformulating feature removal as a differentiable operation where instead of discretely removing features, we replace a proportion of each feature."], "abstract": "Instance-wise feature selection is a valuable tool for interpreting labeled data and the predictions of black-box models. In contrast to global feature selection techniques, instance-wise methods dynamically identify important features for each instance. A growing number of methods learn a selector, which identifies important features, and a predictor, which uses these to make predictions. However, these pioneering methods face challenges including information leakage and lack of differentiability, which can slow training. In this paper, we present Hide&Seek, an end-to-end differentiable model for instance-wise feature selection. We jointly learn feature selection and prediction under a single objective without information leakage. Hide&Seek outperforms existing state-of-the-art models across a range of experiments and is fast to train. We achieve this by reformulating feature removal as a differentiable operation where instead of discretely removing features, we replace a proportion of each feature. Training is further stabilized via a parsimony-weight annealing framework.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 77726, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Interpretability", "Memory / RAG", "Deep Research", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["O1L", "O1", "R1", "phic", "phinyanaphongs", "ying", "yields", "T5"], "matched_methods": [], "checked_on": "2026-08-19T03:32:30.410713+00:00"}}
{"id": "evg-6e4795077e1c553d", "title": "HarnessEval-W: Agentifying the Evaluation of Visual Worlds", "authors": ["Weiliang Chen", "Haowen Sun", "Jun Gao", "Jiawei Chi", "Hanyang Wang", "Qiyu Dai", "Yihao Li", "Hao Li", "Jingnan Gao", "Yi-Hsin Hung", "Xingzhuo Guo", "Shangchen Miao", "Zhiyuan Shi", "Xiang Li", "Fengrui Tian", "Weihua Du", "Ziqi Huang", "Shenyuan Gao", "Siqiao Huang", "Mingyu Liu", "Yifei Li", "Shizun Wang", "Xi Wang", "Tianqi Zhang", "Xue Luo", "Xiyin Ren", "Jinshan Ren", "Xiaoyang Shen", "Xiaobo Hu", "Zhiyang Dou", "Mingyu Ding", "Yichao Yan", "Xinchao Wang", "Yizhou Wang", "Shilong Liu", "Wenzhao Zheng", "Yueqi Duan", "Yuan Gong", "Ziwei Liu", "Ming-Yu Liu", "Jialong Wu", "Jiangran Lyu", "Fangfu Liu"], "year": 2026, "published": "2026-08-17T17:43:24Z", "updated": "2026-08-17T17:43:24Z", "arxiv_id": "http://arxiv.org/abs/2608.16859v1", "url": "https://arxiv.org/abs/2608.16859v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["World Model"], "benchmarks": [], "models": ["ying"], "key_results": ["A benchmark should deliver more than a scalar score: what makes an evaluation trustworthy is the reasoning that justifies the score."], "abstract": "A benchmark should deliver more than a scalar score: what makes an evaluation trustworthy is the reasoning that justifies the score. This is especially critical for world models, where judging a rollout requires understanding whether physics, causality, and world state evolve correctly. Humans spot such violations naturally, yet no existing benchmark automates this capability: metrics are computed brute-force, leaving no reasoning chain that can be examined or verified. We introduce HarnessEval-W, an agentified evaluation pipeline that brings the harness paradigm from the LLM ecosystem to world model benchmarking. Rather than applying a fixed rubric, HarnessEval-W interprets the context of each evaluation case, decomposes the evaluation question into measurable subproblems, and spawns specialized sub-agents, each equipped with tailored context and diagnostic tools to reason over its own subproblem. The parent agent then validates the gathered evidence and summarizes it into the final verdict. This hierarchical workflow turns every evaluation into a transparent evidence tree whose complete reasoning chain justifies the result. We apply HarnessEval-W to 18 representative world models over 330 evaluation cases. Its judgments closely align with human preferences while providing verifiable, fine-grained diagnoses of every generated rollout. We open-source the full pipeline as a live benchmark and invite the broad community to contribute to grow new skills and evaluation cases as world models evolve.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 71725, "fulltext_methods": ["Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Long Context", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "Tool Use", "Memory / RAG", "Computer Use", "Deep Research"], "fulltext_benchmarks": ["MT-Bench", "Chatbot Arena"], "fulltext_models": ["GPT-5.5", "Grok Imagine 1.5", "Grok Imagine Video 1.5. Note", "ying", "yields", "Yin", "Ying", "Yi", "FLUX 3", "FLUX 3 Video"], "matched_methods": ["World Model"], "checked_on": "2026-08-19T03:32:50.427169+00:00"}}
{"id": "evg-66003dd973c5de39", "title": "Diagnosing Dense Same-Class Attribute Misbinding in Large Vision-Language Models", "authors": ["Yuanzhi Xu", "Qian Gao", "Jun Fan", "Guohui Ding", "Zhenyu Yang", "Yuteng Xiao", "Sixue Lin"], "year": 2026, "published": "2026-08-17T17:01:21Z", "updated": "2026-08-17T17:01:21Z", "arxiv_id": "http://arxiv.org/abs/2608.16805v1", "url": "https://arxiv.org/abs/2608.16805v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["yielding"], "key_results": ["Large vision-language models can recognize the objects and attributes in a crowded scene yet assign an attribute to the wrong same-class instance."], "abstract": "Large vision-language models can recognize the objects and attributes in a crowded scene yet assign an attribute to the wrong same-class instance. Generic visual-question-answering accuracy marks the response as wrong, while object-hallucination metrics may regard both the object and attribute as image-supported; neither reveals the transfer. This study formalizes this blind spot as Dense Same-Class Attribute Misbinding (DSCAM) and presents InstaBind-Lite, a controlled benchmark that makes it directly measurable. Its 524 images contain 529 curated groups of 3-6 same-class entities, 1773 boxed instances, ordered neighbors, distinguishable color-like attributes, and four complementary question levels, yielding 9580 deterministically evaluated questions. Unlike existing protocols, source-instance annotations separate unsupported generation and recognition failure from an attribute copied from another visible entity. Binding-specific metrics further quantify transfer frequency, adjacency, ordinal distance, and intervention effects. Across five open-source and two commercial/API models, the open-source systems average 19.84% Misbinding Rate and the API systems 7.55%; these errors are hidden by aggregate accuracy. Among identifiable transfers, 80.70% and 81.51%, respectively, originate from adjacent instances. Localization and instance-first interventions help selected models but are not universal remedies. InstaBind-Lite therefore turns previously undifferentiated wrong answers into source-identifiable failure categories and tests a reliability dimension that conventional benchmarks cannot determine: whether a model knows not only what is visible, but which instance owns each attribute.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 38497, "fulltext_methods": ["Test-time Scaling", "Interpretability", "Safety / Jailbreak", "VLM"], "fulltext_benchmarks": ["COCO"], "fulltext_models": ["Gemini-3.5-Flash", "Gemini-3.5-Flash API 80.13 5.79 29.15 78.38 9.66 1.42 Qwen3-VL-Plus API 81.52 9.31 50.4 84.64 7.53 0.02 Table III demonstrates the diagnostic information added by InstaBind-Lite. Aggregate accuracy identifies Qwen3-VL-Plus and Gemini-3.5-Flash as the strongest evaluated systems", "Gemini-3.5-Flash 80.13", "Gemini-3.5-Flash remains distinctly low at 5.79", "Gemini", "Gemini-3.5-Flash 71.46 7.16 71.53 8.97 70.55 9.18 Qwen3-VL-Plus 74.23 8.89 71.84 10.02 72.64 11.24 Table V separates models whose errors are sensitive to visual competition from those that need broader context. LLaVA-1.5 gains 20.14 accuracy points and reduces MBR by 19.12 points under crop oracle", "Gemini-3.5-Flash and Qwen3-VL-Plus begin with lower full-image MBR but increase under localized views. Proprietary preprocessing prevents separation of global-context removal from crop-induced scale or distribution shift. The negative gaps therefore do not imply that visual competition is beneficial", "Gemini -1.81", "Qwen2.5-VL-7B", "Qwen3-VL-Plus"], "matched_methods": ["VLM"], "checked_on": "2026-08-19T03:32:54.250531+00:00"}}
{"id": "evg-68eeb444586ffb9e", "title": "TRACE-Bench: Decomposing and Diagnosing Multi-Reference Image Generation", "authors": ["Haoran Wang", "Chaofan Ma", "Ran Yi", "Lizhuang Ma"], "year": 2026, "published": "2026-08-17T16:15:50Z", "updated": "2026-08-17T16:15:50Z", "arxiv_id": "http://arxiv.org/abs/2608.16765v1", "url": "https://arxiv.org/abs/2608.16765v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Despite recent advances in unified multimodal models for multi-reference image generation, existing benchmarks remain organized around predefined task types (e.g., \"subject composition\"), which are ill-suited to this combinatorial setting and lead to fragmented coverage, uncontrolled complexity, and little diagnostic value."], "abstract": "Despite recent advances in unified multimodal models for multi-reference image generation, existing benchmarks remain organized around predefined task types (e.g., \"subject composition\"), which are ill-suited to this combinatorial setting and lead to fragmented coverage, uncontrolled complexity, and little diagnostic value. Recognizing that diverse multi-reference tasks share a common set of atomic operations, we adopt a capability-oriented perspective and formalize four operators: Anchor ($f$), Disentangle ($g$), Apply ($\\oplus$), and Compose ($C$). Any multi-reference prompt can then be represented as a compositional formula over these operators, whose structural complexity is quantified by the number of operator slots. Building on this formulation, we construct TRACE-Bench, comprising approximately 1,600 evaluation cases across slot counts 1--8, built from 631 formula templates and around 4,000 reference images spanning diverse artistic styles and real-world subjects. The formula structure directly drives an operator-aligned evaluation protocol for per-capability scoring and a diagnostic tree analysis for recursive failure localization. Evaluating 9 leading models reveals insights invisible to holistic scoring: the primary bottleneck lies in disentanglement ($g$) and attribute binding ($\\oplus$) rather than scene-level composition ($C$), with even the best model scoring only 0.74 on attribute fidelity. Project page: https://amuseum-whr.github.io/TraceBench", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 65552, "fulltext_methods": ["RLVR / GRPO", "Long Context", "Preference Optimization", "Safety / Jailbreak", "Multi-Agent", "World Model", "Video Generation", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-5.4", "Gemini-2.5-pro-based tagging pipeline that summarizes foreground entities and their associated attributes in a structured form. Entities are assigned category labels from a predefined ontology", "Gemini-2.5-Pro", "Gemini-2.5-Pro as the VLM judge. We retain a text-only prompt for each case and additionally report text-image similarity computed by CLIP ViT-L", "gemini 2.5 flash image", "gemini-2-5-flash-image", "Qwen-Image-Edit", "Qwen-Image-Edit-2509", "Qwen-Image-Edit-2511", "Qwen"], "matched_methods": ["VLM"], "checked_on": "2026-08-19T03:32:57.586965+00:00"}}
{"id": "evg-610d36b0bdf49740", "title": "GenRouter: Unified Workflow Routing for Agentic Image Generation", "authors": ["Harold Haodong Chen", "Zhiyu Hou", "Wen-Jie Shu", "Weilin Ruan", "Yingjie Xu", "Litao Guo", "Ying-Cong Chen"], "year": 2026, "published": "2026-08-17T15:36:45Z", "updated": "2026-08-17T15:36:45Z", "arxiv_id": "http://arxiv.org/abs/2608.16721v1", "url": "https://arxiv.org/abs/2608.16721v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Extensive experiments across diverse benchmarks demonstrate that GenRouter achieves superior visual alignment while reducing execution costs by over 95% and latency by 65% compared to heavyweight static pipelines.", "Furthermore, the system continuously self-evolves via accumulated experience, enabling robust zero-shot generalization that boosts performance and halves computational overhead."], "abstract": "The rapid evolution of text-to-image (T2I) generation models has effectively solved the foundational challenge of raw pixel synthesis, shifting the community's focus toward fulfilling increasingly intricate user requests. While recent agentic image generation workflows enhance static inference with advanced capabilities like external knowledge retrieval and iterative reasoning, they mostly operate in isolated silos with fixed ``one-size-fits-all\" topologies. This inevitably leads to severe compute-mismatch, where simple queries are forced through computationally heavy pipelines. To bridge this gap, we present GenRouter, the first unified workflow routing framework for agentic image generation. We first formulate GenCanvas, standardizing diverse agentic pipelines into a universal set of foundational primitives and executable templates. Operating over this unified space, GenRouter adaptively routes heterogeneous prompts to their optimal workflows via (i) demand profiling, (ii) experience matching, and (iii) Pareto filtering. Extensive experiments across diverse benchmarks demonstrate that GenRouter achieves superior visual alignment while reducing execution costs by over 95% and latency by 65% compared to heavyweight static pipelines. Furthermore, the system continuously self-evolves via accumulated experience, enabling robust zero-shot generalization that boosts performance and halves computational overhead.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 62211, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Distillation", "Preference Optimization", "Interpretability", "Multi-Agent", "Tool Use", "Memory / RAG", "Video Generation", "VLM"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["gpt-4v", "o1", "Qwen-Image", "Qwen3.5-4B", "Qwen-Image-2512", "Qwen-Image-Edit-2511", "Qwen-image", "Qwen3", "Qwen-image-agent", "r1"], "matched_methods": ["Multi-Agent", "Preference Optimization"], "checked_on": "2026-08-19T03:33:00.793002+00:00"}}
{"id": "evg-df533c53fac9f2ad", "title": "PersonaShot: Benchmarking Person-Centric Narrative Continuity in Multi-Shot Video Generation", "authors": ["Yuji Wang", "Yuheng Chen", "Teng Hu", "Ran Yi", "Yijia Hong", "Han Feng", "Weijian Cao", "Chengjie Wang", "Lizhuang Ma", "Jiangning Zhang"], "year": 2026, "published": "2026-08-17T15:34:54Z", "updated": "2026-08-17T15:34:54Z", "arxiv_id": "http://arxiv.org/abs/2608.16717v1", "url": "https://arxiv.org/abs/2608.16717v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Distillation", "Video Generation", "VLM"], "benchmarks": [], "models": [], "key_results": ["\\textbf{\\textit{3)} Systematic Evaluation and Insights:} Our evaluation reveals distinct capability profiles across state-of-the-art models and a clear gap between perceptual quality and cross-shot narrative continuity."], "abstract": "Video generation is rapidly evolving from single-shot clips to multi-shot narratives, where the human character serves as the core narrative anchor. However, existing benchmarks mainly assess character appearance or individual-shot quality, without measuring whether physical and emotional states remain coherent across cuts. They also rarely provide criterion-specific evaluation methods, although physical continuity, facial dynamics, and cinematic relations require different visual, temporal, and relational evidence. To address these limitations, we introduce PersonaShot, the first person-centric benchmark for narrative continuity in multi-shot video generation. PersonaShot contains approximately 1,000 multi-shot segments and 16 metrics spanning physical continuity, affective dynamics, and cinematic grammar. \\textbf{\\textit{1)} Narrative Continuity Benchmark:} We evaluate character coherence across three temporal levels: within-shot states, cross-shot transitions, and sequence-level trajectories. \\textbf{\\textit{2)} Human-Aligned Specialist Evaluators:} We distill reasoning from a large multimodal teacher into lightweight criterion-specific evaluators, each grounded in the visual, temporal, or relational evidence required by its metric, and align them with expert human judgments. \\textbf{\\textit{3)} Systematic Evaluation and Insights:} Our evaluation reveals distinct capability profiles across state-of-the-art models and a clear gap between perceptual quality and cross-shot narrative continuity. Even visually compelling videos frequently exhibit physical-state resets, abrupt affective shifts, and broken cinematic relations across shots. Human studies further demonstrate strong agreement between our evaluators and expert judgments.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 76294, "fulltext_methods": ["Distillation", "KV Cache", "Preference Optimization", "Interpretability", "Computer Use", "Video Generation", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["GPT-4o-based", "LLaMA 6", "llama", "Qwen-based", "Qwen3.5-397B-A17B", "Qwen3.5-4B", "Qwen", "Qwen3.", "Qwen3-4B", "Qwen3-8B"], "matched_methods": ["Distillation", "VLM", "Video Generation"], "checked_on": "2026-08-19T03:33:03.575870+00:00"}}
{"id": "evg-da5f363730510dd3", "title": "MIRROR: Multimodal Intelligent Radiology Reasoning and Observation Reporter", "authors": ["Vignesh Nagarajan", "Sriram Venkatapathy"], "year": 2026, "published": "2026-08-17T15:25:38Z", "updated": "2026-08-17T15:25:38Z", "arxiv_id": "http://arxiv.org/abs/2608.16709v1", "url": "https://arxiv.org/abs/2608.16709v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["We are precise about what that buys: a MIRROR report's findings are auditable against the probability vector, while the sentences framing them are ordinary generated text, and we show one stating a cardiothoracic ratio the system never measured.", "Under the class imbalance normal in radiology, aggregate metrics flatter models that do nothing, and should be reported against that floor."], "abstract": "A radiologist reading a model's output faces two problems. The model returns a number and no reason, and any system that turns that number into readable prose can quietly add claims the model never made. MIRROR is a research prototype built to separate those failures. It chains a multi-label classifier, a Grad-CAM localizer that turns each positive finding into a named anatomical region, and a report writer that receives the labels, probabilities, and regions but never the image. Because the language layer cannot see pixels, it cannot assert a finding the classifier did not make. We are precise about what that buys: a MIRROR report's findings are auditable against the probability vector, while the sentences framing them are ordinary generated text, and we show one stating a cardiothoracic ratio the system never measured. One registry holds the taxonomy, anatomy, and phrasing for chest X-ray, brain MRI, and head CT, so adding a modality is a data change; all three are routed and tested, one is trained. On ChestMNIST that classifier reaches macro AUROC 0.729 and ranks better than chance on all 14 labels, at 1.6 to 6.8 times the precision a random ranker would get. Yet at the default 0.5 threshold it emits no positive prediction at all for 11 of them, and its excellent-looking Brier score of 0.045 sits beside the 0.047 earned by a predictor that ignores the image. The discrimination is real; the decisions are not. Under the class imbalance normal in radiology, aggregate metrics flatter models that do nothing, and should be reported against that floor.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 44431, "fulltext_methods": ["Interpretability", "Safety / Jailbreak", "Memory / RAG", "Deep Research", "VLM"], "fulltext_benchmarks": ["ImageNet"], "fulltext_models": ["Claude", "phic", "PHI", "ying", "yields"], "matched_methods": ["VLM"], "checked_on": "2026-08-19T03:33:09.567188+00:00"}}
{"id": "evg-37c58c1bd9cca90c", "title": "AnchorScore: A CLIP-Based Diagnostic of MLLM Annotation Difficulty", "authors": ["Yan Ma", "Lizhuo Zhang"], "year": 2026, "published": "2026-08-17T15:10:33Z", "updated": "2026-08-17T15:10:33Z", "arxiv_id": "http://arxiv.org/abs/2608.16690v1", "url": "https://arxiv.org/abs/2608.16690v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["yields"], "key_results": ["Multimodal large language models (MLLMs) are widely used for automated annotation, yet their per-class accuracy varies widely (e.g., 12%-98% across the 13 classes of three classroom sub-datasets) and is expensive to measure: evaluating one 27B MLLM on 5,416 validation images takes roughly 14 hours, whereas a frozen-CLIP pass over the same images completes in about 3 minutes."], "abstract": "Multimodal large language models (MLLMs) are widely used for automated annotation, yet their per-class accuracy varies widely (e.g., 12%-98% across the 13 classes of three classroom sub-datasets) and is expensive to measure: evaluating one 27B MLLM on 5,416 validation images takes roughly 14 hours, whereas a frozen-CLIP pass over the same images completes in about 3 minutes. A low-cost signal for ranking classes by expected MLLM annotation difficulty a priori remains underexplored. Building on the AnchorProxy construct (per-class zero-shot CLIP accuracy) introduced in the companion study, this paper systematically evaluates its full-frame formulation, termed AnchorScore here, as an a priori diagnostic that flags the classes MLLMs are least likely to annotate reliably.\n  On classroom behavior data (SCB5, 13 classes, 6 MLLMs), AnchorScore correlates with per-class MLLM accuracy (Spearman rho = 0.769, p = 0.002, n = 13). None of the alternative difficulty predictors (DINOv2, ResNet-50, SigLIP, or MLLM self-verbalized uncertainty) showed a significant class-level correlation at n = 13. A cross-model consensus control suggests AnchorScore primarily captures a shared class-difficulty factor rather than a CLIP-specific signal. An independent replication on Stanford40 Actions yields a nearly identical effect (rho = 0.817, p < 0.001); the association is strongest on activity-recognition data and attenuates on medical and satellite imagery.\n  Three practical applications follow: a deployable hybrid CLIP/MLLM routing strategy (predicted-class routing: up to +23 pp over CLIP-only at roughly 44% MLLM cost savings), prompt disambiguation on hard classes (exploratory), and review-priority prediction for human verification. AnchorScore does not estimate exact MLLM accuracy; it provides a low-cost ranking signal that directs expensive MLLM evaluation to the classes where it is most informative.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 110944, "fulltext_methods": ["Test-time Scaling", "MoE", "Preference Optimization", "Interpretability", "Multi-Agent", "Computer Use", "Deep Research", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["ImageNet", "AlpacaEval", "MT-Bench", "Chatbot Arena", "DROP"], "fulltext_models": ["GPT-4V", "Gemini Pro", "llama-served", "llama chat", "llama", "llama API", "Qwen2-VL-7B", "Qwen3.5", "Qwen3.6", "Qwen"], "matched_methods": ["VLM"], "checked_on": "2026-08-19T03:33:12.736033+00:00"}}
{"id": "evg-61fa1dc3e33a4859", "title": "Closing the Affective Loop: Multimodal Speaker-Listener Emotion-Dynamics-Aware Empathetic Social Robots", "authors": ["Zi Haur Pang", "Casey Kennington", "Tatsuya Kawahara"], "year": 2026, "published": "2026-08-17T15:08:38Z", "updated": "2026-08-17T15:08:38Z", "arxiv_id": "http://arxiv.org/abs/2608.16686v1", "url": "https://arxiv.org/abs/2608.16686v1", "primary_category": "cs.HC", "categories": ["cs.HC", "cs.CL", "cs.RO"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["These preliminary results suggest that explicitly modeling both speaker emotional dynamics and listener affective state can improve embodied empathetic interaction."], "abstract": "Empathetic social robots should respond not only to what users say, but also to how their emotions dynamically evolve during interaction. However, existing empathetic dialogue systems are often text-centered and primarily model empathy as a one-way mapping from the user's emotion to the system response, limiting their ability to capture embodied speaker--listener affective exchange. We present AffectLoop, a multimodal speaker-listener emotion-dynamics-aware spoken dialogue system implemented on the Misty II robot. The system tracks the speaker's verbal and facial affective dynamics, estimates the robot listener's own verbal and behavioral affective state, and conditions LLM-based response generation on both affective streams. The robot then generates a short spoken empathetic response together with emotionally congruent embodied behavior, forming a closed speaker--listener affective loop. We evaluate the system in a pilot within-subject study with five participants, comparing it with an otherwise identical utterance-conditioned baseline that omits the speaker- and listener-affective-state inputs. The proposed system received higher overall impression ratings, especially for empathetic response and user satisfaction. Post-hoc log analysis further showed higher speaker-listener affective alignment and stronger valence-based distress recovery. These preliminary results suggest that explicitly modeling both speaker emotional dynamics and listener affective state can improve embodied empathetic interaction.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 29878, "fulltext_methods": ["Preference Optimization", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-4.1-nano22", "gpt-4-1", "ying"], "matched_methods": ["Preference Optimization", "VLM"], "checked_on": "2026-08-19T03:33:16.645618+00:00"}}
{"id": "evg-0bb37271e46e6e20", "title": "HarmTrace: Anchor-Calibrated Decoupled Optimization for Fine-Grained Target Identification in Harmful Memes", "authors": ["Yujia Li", "Yiqun Zhang", "Zihan Cheng", "Yijie Huang", "Tenglong Ye", "Zihan Wang", "Xiaocui Yang", "Shi Feng", "Yifei Zhang", "Daling Wang"], "year": 2026, "published": "2026-08-17T14:21:41Z", "updated": "2026-08-17T14:21:41Z", "arxiv_id": "http://arxiv.org/abs/2608.16622v1", "url": "https://arxiv.org/abs/2608.16622v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["RLVR / GRPO", "Safety / Jailbreak", "VLM"], "benchmarks": [], "models": ["Qwen3-VL-8B", "ying"], "key_results": ["HarmTrace improves both JRA and harmfulness accuracy across the evaluated backbones, with JRA on the Qwen3-VL-8B backbone increasing from 17.58\\% to 52.51\\%."], "abstract": "Multimodal harmful meme detection is typically formulated as image--text harmfulness classification. A model may correctly predict harmfulness while misidentifying the attacked target or its supporting evidence. We therefore extend harmful meme detection with fine-grained target identification, asking what type of target is attacked, who is targeted, and where the target appears in the meme. The model predicts harmfulness for every meme and, for harmful memes, outputs the target category, target entity, textual mention, and visual region. To support this task, we introduce Meme3W, which unifies multiple public harmful meme datasets and provides human-verified annotations for harmful instances. We further introduce Joint Record Accuracy (JRA), a strict record-level metric requiring the harmfulness label and all target-identification fields to be jointly correct. Experiments with representative multimodal large language models reveal a substantial gap between harmfulness accuracy and JRA. To narrow this gap, we propose HarmTrace, an anchor-calibrated decoupled optimization framework. HarmTrace strengthens target-entity supervision through entity-aware supervised fine-tuning. It then applies Conditional Target-identification Policy Optimization (CTPO) to decouple harmfulness and target-identification advantages, restricting target-identification optimization to label-correct responses for harmful examples. CTPO uses a Virtual Positive Anchor (VPA) as a fully correct reference for target-identification advantage normalization. HarmTrace improves both JRA and harmfulness accuracy across the evaluated backbones, with JRA on the Qwen3-VL-8B backbone increasing from 17.58\\% to 52.51\\%. Our code is publicly available at https://github.com/llly1234/HarmTrace-for-Harmful-Memes.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 61664, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Distillation", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Multi-Agent", "Computer Use", "VLM", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-5.2", "GPT-5", "GPT-5.2.", "Gemini-3-Flash", "Gemini-3-Flash to generate one explain-then-answer trajectory for each training meme . Both trajectory generation and filtering use only the training split. We retain only trajectories whose structured answers are valid and consistent with the verified gold annotations", "Gemini 3 Flash", "Gemini3-Flash 74.88 73.66 47.95 79.91 73.29 61.87 63.01 65.98 59.59 GPT-5.2 75.92 69.01 28.31 62.56 60.73 43.38 46.80 38.58 33.33 GLM-5V-Turbo 73.38 72.28 44.75 74.89 68.26 60.96 62.56 62.79 57.31 Open-source Large MLLMs Kimi-K2.5 78.27 75.86 38.81 76.26 73.06 55.48 56.85 59.59 55.02 Gemma-4-26B-A4B 78.46 74.24 28.54 67.81 57.08 52.97 54.11 62.79 41.78 Qwen3.5-397B-A17B 75.82 74.58 38.13 75.11 70.78 59.36 63.70 73.06 59.13 Qwen3-VL-235B-A22B 76.01 70.99 21.46 63.47 52.97 37.44 39.50 56.39 47.49 Qwen3.6-27B 76.95 73.80 36.99 69.63 61.87 53.42 55.25 63.24 59.82 Qwen3.5-27B 75.16 73.65 35.84 74.89 65.98 52.05 53.65 63.93 59.36 Qwen3-VL-32B 75.26 73.51 29.00 76.03 67.35 50.91 52.05 65.98 50.68 Open-source Small MLLMs EXPO-HM", "gemini research and technology. arXiv", "gemini", "gemini-3-flash"], "matched_methods": ["RLVR / GRPO", "Safety / Jailbreak", "VLM"], "checked_on": "2026-08-19T03:33:21.225799+00:00"}}
{"id": "evg-6abfb1b23bb1ffb0", "title": "SQuad: Sub-Quadratic Attention Distillation for Efficient Video Generation", "authors": ["Animesh Karnewar", "Denis Korzhenkov", "Amirhossein Habibian", "Mohsen Ghafoorian"], "year": 2026, "published": "2026-08-17T13:47:44Z", "updated": "2026-08-17T13:47:44Z", "arxiv_id": "http://arxiv.org/abs/2608.16585v1", "url": "https://arxiv.org/abs/2608.16585v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Distillation", "Preference Optimization", "Memory / RAG", "Video Generation"], "benchmarks": ["SQuAD"], "models": [], "key_results": ["Motivated by this, we propose SQuad, a Sub-Quadratic Attention Distillation framework that achieves a complexity of $\\mathcal{O}(n\\sqrt{n})$ in the resulting distilled Attention, naturally balancing the efficiency v/s expressivity trade-off.", "Instead of training our own Video DiT from scratch, which is prohibitively expensive, we fit a pretrained full softmax Self-Attention DiT into our proposed SQuad-Attention one by distilling the former in two stages: Flow-Matching Supervised Fine-Tuning (SFT), followed by improved Distribution Matching Distillation (DMD2) which additionally makes the sampling more efficient.", "On the Wan~2.2 5B text-to-video model, SQuAD matches the quadratic teacher on VBench ($83.20$ v/s $83.08$) while cutting the per-step per-block attention FLOPs by $\\sim$$67\\times$ and attention latency by $\\sim$$11\\times$, and end-to-end DiT latency by 2$\\times$, all while also generating a video in only $6$ Neural Functional Evaluations (NFEs) instead of the default $100$."], "abstract": "Video Diffusion Transformers (DiTs) spend most of their compute inside the Self-Attention operation, whose cost grows quadratically, $\\mathcal{O}(n^2)$, with the number of latent tokens $n$. For the task of video generation, the token count is large, so this term dominates runtime and memory, and thereby caps the resolution and duration we can generate. Linear $\\mathcal{O}(n)$ and low-rank $\\mathcal{O}(nk)$ surrogates of Self-Attention trade the full softmax $QK^T$ for cheaper kernels, but rarely recover the original's expressivity, leaving a stubborn quality gap. Motivated by this, we propose SQuad, a Sub-Quadratic Attention Distillation framework that achieves a complexity of $\\mathcal{O}(n\\sqrt{n})$ in the resulting distilled Attention, naturally balancing the efficiency v/s expressivity trade-off. Instead of training our own Video DiT from scratch, which is prohibitively expensive, we fit a pretrained full softmax Self-Attention DiT into our proposed SQuad-Attention one by distilling the former in two stages: Flow-Matching Supervised Fine-Tuning (SFT), followed by improved Distribution Matching Distillation (DMD2) which additionally makes the sampling more efficient. On the Wan~2.2 5B text-to-video model, SQuAD matches the quadratic teacher on VBench ($83.20$ v/s $83.08$) while cutting the per-step per-block attention FLOPs by $\\sim$$67\\times$ and attention latency by $\\sim$$11\\times$, and end-to-end DiT latency by 2$\\times$, all while also generating a video in only $6$ Neural Functional Evaluations (NFEs) instead of the default $100$.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": false, "verification": {"source": "unavailable", "status": "unavailable", "checked_on": "2026-08-19T03:33:23.898272+00:00"}}
{"id": "evg-a6a23d3dd800c218", "title": "DSPrompt: Dynamic Soft Prompt Defense Against M-RAG Corruption", "authors": ["Chang Liu", "Yuni Lai", "Mingyue Cui", "Cong Tian", "Yunyan Zhang", "Xian Wu", "Kai Zhou", "Bin Xiao"], "year": 2026, "published": "2026-08-17T13:11:51Z", "updated": "2026-08-17T13:11:51Z", "arxiv_id": "http://arxiv.org/abs/2608.16536v1", "url": "https://arxiv.org/abs/2608.16536v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.CL"], "pillar": "Multimodal / World Models", "methods": ["Safety / Jailbreak", "Memory / RAG", "VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["These prompts are trained under a dynamic min-max scheme: an online multimodal attacker continually crafts hard adversarial documents against the current retriever, while the defender is updated to push such documents out of the top-k while preserving the ranking and diversity of benign evidence.", "Extensive experiments across four benchmarks and three representative poisoning attacks show that DSPrompt substantially reduces the attack success rate and poison retrieval rate while maintaining near-lossless retrieval utility and generation fidelity, consistently outperforming existing defense baselines at a fraction of their computational cost."], "abstract": "Multimodal Retrieval Augmented Generation (M-RAG) is increasingly vulnerable to adversarial attacks where malicious data are crafted to produce embeddings that align with benign entries in the vector space, deceiving retrieval and inducing harmful outputs. Existing defenses primarily operate at query time, relying on auxiliary detectors, similarity re-ranking, or feature-consistency checks. However, these approaches suffer from non-trivial inference overhead, generalize poorly to unseen attack strategies, and often assume specific attack distributions. To address this, we propose DSPrompt, a Dynamic Soft Prompt defense framework that directly reshapes the retriever's embedding semantics, without modifying the retrieval pipeline. It inserts few learnable soft prompts into each layer of the visual and textual encoders of a frozen retriever, utilizing a shallow-to-deep length schedule that is adaptive to the capacity in the model layers. These prompts are trained under a dynamic min-max scheme: an online multimodal attacker continually crafts hard adversarial documents against the current retriever, while the defender is updated to push such documents out of the top-k while preserving the ranking and diversity of benign evidence. Because the defended encoder can be pre-computed and indexed exactly as in standard dense retrieval, DSPrompt incurs no additional per-query optimization and introduces fewer than 1% additional parameters. Extensive experiments across four benchmarks and three representative poisoning attacks show that DSPrompt substantially reduces the attack success rate and poison retrieval rate while maintaining near-lossless retrieval utility and generation fidelity, consistently outperforming existing defense baselines at a fraction of their computational cost.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00"}
{"id": "evg-0ca76511ce09d82b", "title": "Matched Outcomes, Divergent Gaze: How Foveated MLLMs Search Compared to Humans", "authors": ["Mohamed Amine Kerkouri", "Marouane Tliba", "Aladine Chetouani", "Ulas Bagci", "Alessandro Bruno"], "year": 2026, "published": "2026-08-17T12:54:39Z", "updated": "2026-08-17T12:54:39Z", "arxiv_id": "http://arxiv.org/abs/2608.16514v1", "url": "https://arxiv.org/abs/2608.16514v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.CL", "cs.HC", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": ["COCO"], "models": [], "key_results": ["Human visual search is serial: the fovea must land on a candidate to confirm it, and those landings form a scanpath."], "abstract": "Human visual search is serial: the fovea must land on a candidate to confirm it, and those landings form a scanpath. Whether multimodal large language models (MLLMs), given the same foveated input, search as humans do bears on their use as models of human vision and on attention-alignment scores. We compare three general-purpose MLLMs with human eye-movement scanpaths on goal-directed search (COCO-Search18), driving each model fixation by fixation through an identical, human-matched foveated view and assessing it along three axes: the decision of target presence, the efficiency of reaching the target, and the gaze process itself. The axes dissociate. On the decision and on target acquisition the models match or exceed humans, detecting present targets near ceiling and reaching them on the first saccade more often than people do. The gaze process is not human. Under the human-matched condition, all three share one signature: low-entropy, large-amplitude, self-consistent scanpaths that agree with themselves far more closely than two humans agree with each other. That is consistent with a single-pass, non-serial architecture rather than a limit of acuity. Matched retinal input reproduces where humans look but not how the looking unfolds in time, and no degradation regime recovers human-like search at human-like success. The gap sits on a process axis that answer-alignment and saliency metrics do not measure. Because they miss it, such metrics cannot certify human-like vision, and zero-shot models suit outcome and spatial questions but not temporal, process-level ones.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00"}
{"id": "evg-d2a5113aba850e9f", "title": "FETERS: Few-Shot Early Time-Series Classification via Effective Ratio Selection", "authors": ["Chen-An Tai", "Yujia Wu", "Vincent S. Tseng"], "year": 2026, "published": "2026-08-17T10:38:31Z", "updated": "2026-08-17T10:38:31Z", "arxiv_id": "http://arxiv.org/abs/2608.16385v1", "url": "https://arxiv.org/abs/2608.16385v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Extensive experiments on 69 public datasets spanning 14 domains show that FETERS achieves state-of-the-art (SOTA) performance in the 5-shot setting, with the highest average harmonic mean (HM) and the best HM on 38 datasets, while outperforming the current SOTA method on 44 datasets."], "abstract": "Early time-series classification (ETSC) aims to make accurate predictions from partially observed time series as early as possible. Although various stopping mechanisms and feature learning strategies have been developed for ETSC, most existing methods assume access to sufficient labeled training data, which may be unrealistic in applications with limited annotation. Under limited supervision, learning an additional sample-level stopping module and extracting effective classification features can both become challenging. In this paper, we propose FETERS, a few-shot ETSC framework that selects a dataset-level stopping ratio through class-wise leave-one-out (LOO) evaluation on the support set and uses a penalty-based reward function to manage the accuracy-earliness trade-off, thereby avoiding the need to train an additional stopping module. FETERS further combines Rocket-based features with frozen Chronos representations for classification. Extensive experiments on 69 public datasets spanning 14 domains show that FETERS achieves state-of-the-art (SOTA) performance in the 5-shot setting, with the highest average harmonic mean (HM) and the best HM on 38 datasets, while outperforming the current SOTA method on 44 datasets. FETERS also remains competitive in the full-shot setting, demonstrating its effectiveness in managing the accuracy-earliness trade-off.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 38656, "fulltext_methods": ["RLVR / GRPO", "Computer Use", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["r1", "yield", "ying", "yields", "yi"], "matched_methods": [], "checked_on": "2026-08-19T03:33:26.535603+00:00"}}
{"id": "evg-bcd3b9ed559fc5db", "title": "Self-Routed Tensor Adapters for Parameter-Efficient Universal Visual Adaptation", "authors": ["Suraj Yadav"], "year": 2026, "published": "2026-08-17T10:35:36Z", "updated": "2026-08-17T10:35:36Z", "arxiv_id": "http://arxiv.org/abs/2608.16384v1", "url": "https://arxiv.org/abs/2608.16384v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Quant × AI", "methods": ["MoE"], "benchmarks": [], "models": [], "key_results": ["MoE-based adapters improve specialization through multiple expert pathways, but often rely on external routers and large expert banks, adding parameters and separating routing from adaptation.", "Across five heterogeneous multi-domain visual classification benchmarks, SRTA achieves competitive or slightly stronger average accuracy than MoE-style PEFT baselines while using substantially fewer trainable parameters."], "abstract": "Universal visual representations require adaptation mechanisms that adapt across heterogeneous domains without fragmenting knowledge into domain-specific modules. Parameter-efficient fine-tuning adapts frozen visual foundation models efficiently, but standard low-rank adapters use a fixed subspace for all inputs, which can be restrictive when domains differ in style, background, and semantic context. MoE-based adapters improve specialization through multiple expert pathways, but often rely on external routers and large expert banks, adding parameters and separating routing from adaptation. We propose \\textbf{Self-Routed Tensor Adapters}, a compact framework for multi-domain visual adaptation. SRTA projects each input into a low-rank space, computes routing weights from this representation using a learnable domain matrix, and uses these weights to blend slices of a shared Tucker core. This produces a sample-specific adaptation matrix without an external gating network, allowing shared visual factors to be reused while supporting domain-aware specialization. To strengthen pathway learning, we introduce a progressive depth-weighted routing objective that supervises routing decisions across adapter layers. Across five heterogeneous multi-domain visual classification benchmarks, SRTA achieves competitive or slightly stronger average accuracy than MoE-style PEFT baselines while using substantially fewer trainable parameters. At rank 64, SRTA uses 2.77M parameters in the 4-domain setting compared with 9.52M for MoLoRA, and 3.00M in the 6-domain setting compared with 14.31M. Overall, SRTA offers an effective accuracy-parameter trade-off for adapting visual foundation models toward universal multi-domain representations. \\href{https://github.com/surajyadav-research/SRTA}{GitHub}", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 37210, "fulltext_methods": ["Test-time Scaling", "MoE", "Interpretability", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["ying", "Yin"], "matched_methods": ["MoE"], "checked_on": "2026-08-19T03:33:29.538951+00:00"}}
{"id": "evg-e327d0acbb64282f", "title": "LaGSplat: Inferring Physics-Governed Interactive Simulation from Monocular Video Using Latent Lagrangian Gaussian Splatting", "authors": ["Louen Pottier"], "year": 2026, "published": "2026-08-17T09:29:12Z", "updated": "2026-08-17T09:29:12Z", "arxiv_id": "http://arxiv.org/abs/2608.16324v1", "url": "https://arxiv.org/abs/2608.16324v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["We present LaGSplat (Latent Lagrangian Gaussian Splatting), a framework that infers interactive, physics-governed dynamics from one or a few monocular videos."], "abstract": "We present LaGSplat (Latent Lagrangian Gaussian Splatting), a framework that infers interactive, physics-governed dynamics from one or a few monocular videos. At inference it lets a user push on the filmed object, rigid or deformable, with an external force that was never measured, annotated, or seen during training. This is possible because a low-dimensional latent state $\\mathbf{q} \\in \\mathbb{R}^d$ plays two roles at once: it is the generalised coordinate of a learned dissipative Lagrangian and the conditioning variable of a Gaussian Splatting decoder. The inductive bias of this decoder, whose primitives are explicit points $μ_i(\\mathbf{q})$ that move with the object, is what lets a force $f$ applied in the image pull back into a latent generalised force $J(\\mathbf{q})^\\top f$ and enter the equations of motion, which pixel-space (CNN) or neural-field (NeRF) decoders cannot do. We validate LaGSplat on test cases of increasing difficulty, from rigid to deformable and from autonomous to forced real systems, combining monocular video and sensor measurements. We further demonstrate interactive use: forces of arbitrary magnitude and direction can be applied to the reconstructed object at any time, its response rendered in real time, in 2D or 3D. Assuming a dissipative Euler-Lagrange equation over a few generalised coordinates trades generality for a bounded, plausible response to unseen forces, where an unconstrained predictor diverges.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 121582, "fulltext_methods": ["Preference Optimization", "Interpretability", "Multi-Agent", "World Model", "Video Generation", "Quant / Trading"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["phics", "phism", "ying", "yields", "Yildiz", "Ying", "Yi"], "matched_methods": [], "checked_on": "2026-08-19T03:33:32.949174+00:00"}}
{"id": "evg-c80f4edaaa2000c2", "title": "SAUL: Sharpness-Aware Augmented-Lagrangian Unlearning", "authors": ["Jaewan Choi", "Junyoung Yang", "Sangdon Park"], "year": 2026, "published": "2026-08-17T08:23:05Z", "updated": "2026-08-17T08:23:05Z", "arxiv_id": "http://arxiv.org/abs/2608.16249v1", "url": "https://arxiv.org/abs/2608.16249v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": ["DROP"], "models": ["ying"], "key_results": ["Beyond the complete SAUL framework, we further show on TOFU that applying the augmented-Lagrangian controller as a drop-in modifier to representative baselines improves their post-forgetting utility, demonstrating the practical value of explicit forgetting control."], "abstract": "Machine unlearning in Large Language Models (LLMs) faces a critical trade-off between erasing target knowledge and preserving general utility. We propose SAUL (Sharpness-Aware Augmented-Lagrangian Unlearning), which formulates unlearning as a constrained minimization problem following the principle of \"forget enough, but no more than necessary.\" At its core, SAUL formulates forgetting as an explicit constraint with a prescribed satisfaction criterion, whereas prior unlearning methods typically specify the desired level of forgetting implicitly through optimization objectives. An augmented Lagrangian controller adaptively adjusts forget-side pressure according to constraint violation and can eventually deactivate the forget-side update as the prescribed criterion remains satisfied. Sharpness-aware updates on both retain and forget objectives, together with a dual-optimizer design that maintains role-separated states, further stabilize the resulting unlearning dynamics. We evaluate SAUL on the TOFU, WMDP, and MUSE benchmarks, demonstrating favorable forgetting-utility trade-offs over representative sharpness- and perturbation-based baselines under benchmark-specific forgetting criteria. Beyond the complete SAUL framework, we further show on TOFU that applying the augmented-Lagrangian controller as a drop-in modifier to representative baselines improves their post-forgetting utility, demonstrating the practical value of explicit forgetting control.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 79453, "fulltext_methods": ["Preference Optimization", "Interpretability", "Safety / Jailbreak", "Memory / RAG", "Deep Research"], "fulltext_benchmarks": ["MMLU", "DROP"], "fulltext_models": ["GPT-5.1", "Llama-3.2-1B-Instruct", "Llama-3.2-3B-Instruct", "Llama-2-7B", "LLaMA-3.2-3B", "Llama-2-7B.", "Llama model", "LLaMA-3.2-3B.", "LLaMA-3.2-1B.", "phic"], "matched_methods": [], "checked_on": "2026-08-19T03:33:36.140113+00:00"}}
{"id": "evg-705c9da3c6b3386b", "title": "The Trade-off Between Covariate Dependence and Latent Structure in Representation Learning", "authors": ["Małgorzata Łazęcka", "Ewa Szczurek"], "year": 2026, "published": "2026-08-17T08:20:40Z", "updated": "2026-08-17T08:20:40Z", "arxiv_id": "http://arxiv.org/abs/2608.16245v1", "url": "https://arxiv.org/abs/2608.16245v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Disentangled representation learning seeks latent representations whose indicidual dimensions each align with a distinct covariate."], "abstract": "Disentangled representation learning seeks latent representations whose indicidual dimensions each align with a distinct covariate. Unsupervised approaches typically target latent dimension independence, yet this gives no guarantee that the resulting dimensions align with semantically meaningful covariates. Supervised approaches structure the latent space using observed covariates, but under correlated covariates they cannot simultaneously control one-to-one latent-covariate alignment and latent independence. We introduce a unified, supervised framework that couples latent dimension-covariate dependence with constraints on the latent structure. Within this framework, we show an inherent trade-off, where enforcing latent independence or exclusive one-to-one latent-covariate dependence comes at a provable cost in latent-covariate alignment. We prove that the resulting disentanglement regimes are ordered by the strength of that alignment. Each regime admits a closed-form transformation of the latent space. We apply these transformations post-hoc to realign the representations of pretrained models such as CLIP, DINOv2, and ViT, and we fold them into the inference of informed factor analysis (iFA), a probabilistic model with covariate-informed factors. On simulated and real multi-omics data, we show that both post-hoc alignment and iFA enable controllability of structured latent representations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 72187, "fulltext_methods": ["Distillation", "Preference Optimization", "Interpretability", "Computer Use", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["ImageNet"], "fulltext_models": ["phical", "phica", "Philosophical", "ying", "yield", "yields", "yielding", "bert"], "matched_methods": ["Preference Optimization"], "checked_on": "2026-08-19T03:33:39.450836+00:00"}}
{"id": "evg-1fbb342df67c0228", "title": "REFLEX: Reflexive Equilibrium Fixed-point Learning for Endogenous eXchanges", "authors": ["Vignesh Nagarajan", "Shriraghav Ashok"], "year": 2026, "published": "2026-08-17T06:19:06Z", "updated": "2026-08-17T06:19:06Z", "arxiv_id": "http://arxiv.org/abs/2608.16155v1", "url": "https://arxiv.org/abs/2608.16155v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CE", "cs.GT"], "pillar": "Quant × AI", "methods": ["Safety / Jailbreak", "Quant / Trading"], "benchmarks": [], "models": ["yield"], "key_results": ["In over-the-counter corporate bond markets, dealers compete for client trades by quoting bid and ask prices."], "abstract": "In over-the-counter corporate bond markets, dealers compete for client trades by quoting bid and ask prices. Tighter quotes attract more business, but also informed customers more likely to trade ahead of adverse price moves, leaving the dealer holding the risk. As dealers increasingly use machine learning to set quotes, they retrain these models on the trades their own quotes attract, creating a feedback loop in which each model reshapes the market that generates its next training data. The question is therefore not only whether a quoting model performs well, but whether the market it creates stays stable as the model learns from it. Existing performative prediction theory gives a sharp stability condition, yet expresses it through abstract properties of the learning objective a trading desk cannot measure before deployment. We introduce REFLEX, a framework that replaces those unobservable quantities with three measurable features of dealer behavior: how strongly trading volume responds to tighter quotes, how sharply the dealer's objective bends around its optimum, and how quickly informed flow increases as spreads narrow. REFLEX combines these into a single retraining modulus, a pre-deployment stability margin estimated from a desk's own quote and execution history that predicts whether repeated retraining will converge or amplify itself. In simulation, predicted and measured stability agree within 8%, and competing dealers increase instability by 1.74x with two and 3.16x with three, as predicted. Where ordinary retraining becomes unstable at modulus 1.21, a structurally anchored correction converges as blind retraining collapses. Calibrated over 36 years of public market data, stability headroom falls roughly 4.4x for investment grade and 4.3x for high yield from calm to crisis regimes. Ultimately, REFLEX turns an abstract convergence theorem into a market-level safety margin.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 33148, "fulltext_methods": ["Safety / Jailbreak", "Multi-Agent", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["R1", "yield", "ying", "yields"], "matched_methods": ["Quant / Trading", "Safety / Jailbreak"], "checked_on": "2026-08-19T03:33:44.697372+00:00"}}
{"id": "evg-2a27cb2f754fca2b", "title": "AdROD: HyperNetwork-based Adversarially Robust Object Detection for Autonomous Driving", "authors": ["Yuting Wu", "Dongfang Guo", "Xiangzhong Luo", "Qun Song", "Rui Tan"], "year": 2026, "published": "2026-08-17T02:52:01Z", "updated": "2026-08-17T02:52:01Z", "arxiv_id": "http://arxiv.org/abs/2608.16031v1", "url": "https://arxiv.org/abs/2608.16031v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "Quant × AI", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["To further improve adversarial robustness, AdROD incorporates a novel \\emph{functional diversity} mechanism, which couples stochastic weight updates with unique input-space transformations.", "Through comprehensive evaluation with synthetic benchmarks, physically deployed adversarial patches, and end-to-end safety tests in the OpenCDA co-simulator, AdROD outperforms five baseline defenses and exhibits superior generalizability compared with the evaluated adversarial-training baselines, while maintaining real-time performance for safely stopping the vehicle at a stop sign instrumented with adversarial patches."], "abstract": "Camera-based object detectors are vulnerable to physical adversarial attacks designed to suppress detections. While adversarial training and input purification offer some protection, they often overfit to specific attack distributions and fail on adaptive adversaries. This paper presents AdROD, an embedded, stochastic ensemble defense software designed for autonomous driving. AdROD employs {\\em low-rank HyperNetworks}, which require only 1.6\\% of the parameter footprint of standard HyperNetworks, to generate diverse detectors at a per-frame rate, making it impractical for attackers to obtain the deployed detectors in time. To further improve adversarial robustness, AdROD incorporates a novel \\emph{functional diversity} mechanism, which couples stochastic weight updates with unique input-space transformations. We design two serving modes of AdROD that strike different trade-offs between robustness and runtime overhead: AdROD-I, a continuous protection mode for maximum resilience that leverages inter-detector disagreement to recover compromised detections, and AdROD-II, an on-demand mode triggered by kinematic discontinuities in object tracking. Through comprehensive evaluation with synthetic benchmarks, physically deployed adversarial patches, and end-to-end safety tests in the OpenCDA co-simulator, AdROD outperforms five baseline defenses and exhibits superior generalizability compared with the evaluated adversarial-training baselines, while maintaining real-time performance for safely stopping the vehicle at a stop sign instrumented with adversarial patches.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 80428, "fulltext_methods": ["RLVR / GRPO", "Safety / Jailbreak", "Memory / RAG", "Deep Research", "World Model", "Quant / Trading"], "fulltext_benchmarks": ["ImageNet", "COCO", "DROP"], "fulltext_models": ["V3", "phic", "yielding", "ying", "yields", "yield"], "matched_methods": ["Safety / Jailbreak"], "checked_on": "2026-08-19T03:33:47.021329+00:00"}}
{"id": "evg-6cf320011b4e05e1", "title": "The Null Token Knows: Reducing Message-Free Hallucination in ASR and NMT", "authors": ["Kirill Borodin", "Vasiliy Kudryavtsev", "Ivan Viakhirev"], "year": 2026, "published": "2026-08-16T21:49:54Z", "updated": "2026-08-16T21:49:54Z", "arxiv_id": "http://arxiv.org/abs/2608.15940v1", "url": "https://arxiv.org/abs/2608.15940v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG", "cs.SD"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Modern encoder-decoder systems can produce fluent text even when their input contains no recoverable message."], "abstract": "Modern encoder-decoder systems can produce fluent text even when their input contains no recoverable message. We study this failure in ASR and NMT through the models' reserved null tokens, asking whether the score for ending generation already carries a usable abstention signal. Across speech recognizers and translation models, we audit native null-token scores and scalar logit shifts. In Whisper, we additionally probe decoder states and compare supervised row edits with conventional external gates. The evaluated models often expose a useful abstention signal, but stock decoding does not reliably act on it. Raising the null-token score can sharply suppress fabrication, but aggressive intervention also deletes valid speech or shortens legitimate translations. These findings turn the null token into a diagnostic lens on hallucination and motivate evaluating abstention methods by both suppression and deletion costs, rather than by hallucination reduction alone.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 70179, "fulltext_methods": ["Search / MCTS", "Test-time Scaling", "Distillation", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Computer Use", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["v3.1", "v3", "v3.", "ying", "BERT"], "matched_methods": ["Quant / Trading"], "checked_on": "2026-08-19T03:33:51.450683+00:00"}}
{"id": "evg-38e6ed1e860490cc", "title": "Resource-Efficient QUBO Formulation for Anchored Currency Arbitrage", "authors": ["Eric A. F. Reinhardt", "Adam J. Hauser"], "year": 2026, "published": "2026-08-16T18:47:57Z", "updated": "2026-08-16T18:47:57Z", "arxiv_id": "http://arxiv.org/abs/2608.15889v1", "url": "https://arxiv.org/abs/2608.15889v1", "primary_category": "quant-ph", "categories": ["quant-ph", "cs.LG"], "pillar": "Quant × AI", "methods": ["Deep Research", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["We demonstrate the efficacy of this formulation using classical simulated annealing against an exact Held-Karp baseline on the same CPU and show that it can effectively find profitable cycles and account for trading fees."], "abstract": "Currency arbitrage (CA) involves trading currencies in cycles to exploit discrepancies in market valuations. Quadratic unconstrained binary optimization (QUBO) involves minimizing a quadratic cost (energy) function of binary variables. Previous works have explored the use of QUBO to solve CA problems. We build on these previous works by introducing realistic constraints such as beginning cycles from a held currency and accounting for per-transaction trading fees. We show that this formulation requires fewer logical variables (qubits) than previous QUBO encodings in the literature. We derive provably sufficient penalty weights for its constraint terms. We also introduce an exact anchor-gauge reweighting of the exchange rates that compresses the QUBO coefficient range from the rate scale to the arbitrage scale, addressing the finite analog precision of annealing hardware. We demonstrate the efficacy of this formulation using classical simulated annealing against an exact Held-Karp baseline on the same CPU and show that it can effectively find profitable cycles and account for trading fees. Finally, we benchmark faithful implementations of five prior QUBO encodings at matched sampler budgets and show that the proposed encoding is the only one to recover the exact fee-adjusted optimum.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 51925, "fulltext_methods": ["Memory / RAG", "Deep Research", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["Claude Opus 4.8 was used for simulated review", "Claude Opus 4.6-5.0 was used as a coding assistant for debugging", "ying", "yields"], "matched_methods": ["Deep Research", "Quant / Trading"], "checked_on": "2026-08-19T03:33:53.915719+00:00"}}
{"id": "evg-e9389ecae46e722a", "title": "Self-Supervised Auxiliary Task Discovery for Stable Reinforcement Learning in Stock Trading", "authors": ["Arishi Orra", "Himanshu Choudhary", "Manoj Thakur"], "year": 2026, "published": "2026-08-16T16:20:40Z", "updated": "2026-08-16T16:20:40Z", "arxiv_id": "http://arxiv.org/abs/2608.15841v1", "url": "https://arxiv.org/abs/2608.15841v1", "primary_category": "cs.LG", "categories": ["cs.LG", "q-fin.CP", "stat.ML"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Auxiliary tasks are often used to improve representation learning and stabilize training, yet they are usually designed manually and depend heavily on prior assumptions about targets and prediction horizons.", "These tasks are updated using a meta gradient mechanism that accounts for their long-term impact on trading performance and improves training stability.", "The empirical results demonstrate that automatically discovered auxiliary tasks lead to more robust learning and improved trading performance compared to existing baselines."], "abstract": "Reinforcement learning has gained increasing attention as a data-driven approach for stock trading. However, learning a policy that is both profitable and stable remains challenging due to non-stationary market behaviour and noisy reward signals. Auxiliary tasks are often used to improve representation learning and stabilize training, yet they are usually designed manually and depend heavily on prior assumptions about targets and prediction horizons. Such fixed designs may not remain suitable across changing market regimes. In this work, we propose a self-supervised framework that automatically discovers auxiliary tasks to support reinforcement learning for stock trading. The auxiliary tasks are formulated as General Value Functions so that their predictions enrich the learned state representation and assist policy optimization. The framework consists of two networks. The main network learns the trading policy along with the auxiliary predictions, while the secondary network generates the definitions of auxiliary tasks through learned cumulants and discount factors. These tasks are updated using a meta gradient mechanism that accounts for their long-term impact on trading performance and improves training stability. We evaluate the proposed approach across four major equity indices: DJI, FTSE, Sensex, and TAIEX. The empirical results demonstrate that automatically discovered auxiliary tasks lead to more robust learning and improved trading performance compared to existing baselines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 72144, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Memory / RAG", "Quant / Trading"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["ying", "yield", "yields", "yielding", "bert"], "matched_methods": ["Quant / Trading", "RLVR / GRPO"], "checked_on": "2026-08-19T03:34:04.858401+00:00"}}
{"id": "evg-3ea190f18471639e", "title": "Learning Stock Trading Policies via Barycenter-Based Adversarial Inverse Reinforcement Learning", "authors": ["Arishi Orra", "Himanshu Choudhary", "Manoj Thakur"], "year": 2026, "published": "2026-08-16T14:45:28Z", "updated": "2026-08-16T14:45:28Z", "arxiv_id": "http://arxiv.org/abs/2608.15770v1", "url": "https://arxiv.org/abs/2608.15770v1", "primary_category": "cs.LG", "categories": ["cs.LG", "stat.ML"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Quant / Trading"], "benchmarks": [], "models": ["yielding"], "key_results": ["Across all the markets, the proposed approach achieves stronger performance than both classical trading rules and recent deep reinforcement learning methods, while exhibiting more stable risk characteristics."], "abstract": "Designing effective trading strategies using reinforcement learning remains challenging due to delayed and noisy rewards, poor exploration, and the difficulty of enforcing explicit risk constraints. In this work, we propose BRaG, a barycenter-based adversarial inverse reinforcement learning framework for stock trading that learns trading behavior from multiple heterogeneous expert strategies. BRaG aggregates expert demonstrations using a performance-weighted Wasserstein barycenter, yielding a stable pseudo-expert representation that captures shared structure across diverse trading styles. This representation is used to pretrain a trading policy via adversarial imitation learning, which alleviates unstable exploration during reinforcement learning. The pretrained policy is subsequently refined using reinforcement learning with true market rewards. To ensure risk-aware decision-making, BRaG incorporates control barrier functions that constrain action execution and regularize policy learning to satisfy drawdown limits. We evaluate the proposed approach on four major global equity markets, including the US, UK, Indian, and Taiwanese indices. Across all the markets, the proposed approach achieves stronger performance than both classical trading rules and recent deep reinforcement learning methods, while exhibiting more stable risk characteristics.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00"}
{"id": "evg-dec1da78ae75e85f", "title": "A contribution to the critique of blockchain censorship", "authors": ["Ruichao Jiang", "Michelle Yeo", "Long Wen"], "year": 2026, "published": "2026-08-16T09:11:09Z", "updated": "2026-08-16T09:11:09Z", "arxiv_id": "http://arxiv.org/abs/2608.15640v1", "url": "https://arxiv.org/abs/2608.15640v1", "primary_category": "cs.CR", "categories": ["cs.CR", "q-fin.TR"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["We study the blockchain censorship attack introduced in [21], which shows that joining the attack is a dominant strategy."], "abstract": "We study the blockchain censorship attack introduced in [21], which shows that joining the attack is a dominant strategy. We show that, by introducing certain detectability threshold, joining the attack can lead to strictly less reward for whales, which are defined to be a small number of validators that hold significantly more voting power than the rest (henceforth known as minnows). This leads to a change of the equilibrium: With whales unwilling to participate in the attack, it is difficult for minnows alone to launch the attack. We also perform Monte Carlo simulation to show the existence of reduction for whales' reward in Ethereum and Solana.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:31:38.137186+00:00"}
{"id": "evg-073a9645f5e48bec", "title": "GoalEvolve: From Handcrafted Algorithm Priors to Goal-Driven Evolution of Physical Design Algorithms", "authors": ["Haixu Liu", "Lei Zhou", "Yuhao Ren", "Yumao Wu", "Zhiang Wang"], "year": 2026, "published": "2026-08-17T15:45:33Z", "updated": "2026-08-17T15:45:33Z", "arxiv_id": "http://arxiv.org/abs/2608.16733v1", "url": "https://arxiv.org/abs/2608.16733v1", "primary_category": "cs.AR", "categories": ["cs.AR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["Physical design algorithms operate within tightly coupled, multi-stage optimization flows, where stage-local gains may vanish or induce downstream degradation.", "Across eight ASAP7 designs, GoalEvolve improves post-route TNS by 30.67% on average and reduces leakage and dynamic power by 21.18% and 9.42% versus default OpenROAD.", "Relative to commercial-tool goals, it closes 62.20% of the normalized power gap on power-dominant designs, surpasses the TNS goals on both timing-dominant designs, and closes 32.48% of the equal-weight timing-power gap on joint designs."], "abstract": "Physical design algorithms operate within tightly coupled, multi-stage optimization flows, where stage-local gains may vanish or induce downstream degradation. Existing program-evolution frameworks often rely on stage-local objectives or undifferentiated multi-metric feedback, which neither guarantee better final results nor identify which unmet requirement should guide the next iteration. We present GoalEvolve, a goal-driven framework that makes physical design algorithm evolution accountable for the final quality of results (QoR) of the complete flow. Given a multi-objective QoR target region, GoalEvolve converts unmet requirements into normalized target gaps, identifies the dominant bottleneck, and uses stage-resolved checkpoint evidence to locate the responsible stage. An LLM-based Teacher then narrows the search to a relevant algorithmic decision and source region, while parallel Student agents implement and validate hypotheses through full-flow evaluation. Local effects, optimization debt, and downstream retention are retained as mechanism evidence for subsequent evolution. Across eight ASAP7 designs, GoalEvolve improves post-route TNS by 30.67% on average and reduces leakage and dynamic power by 21.18% and 9.42% versus default OpenROAD. Relative to commercial-tool goals, it closes 62.20% of the normalized power gap on power-dominant designs, surpasses the TNS goals on both timing-dominant designs, and closes 32.48% of the equal-weight timing-power gap on joint designs. Across all three designs evaluated against Codex goal mode under matched budgets, GoalEvolve further improves TNS by 26.46% while reducing leakage and dynamic power by 12.38% and 0.76%, respectively.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:32:03.905232+00:00", "verified": false, "verification": {"source": "unavailable", "status": "unavailable", "checked_on": "2026-08-19T03:31:12.907390+00:00"}}
{"id": "evg-0a798dc1d3d3f67c", "title": "Chronocooked: A Benchmark for Implicit Interval Timing in Reinforcement Learning Agents", "authors": ["Amrapali Pednekar", "Alvaro Garrido-Perez", "Yara Khaluf", "Pieter Simoens"], "year": 2026, "published": "2026-08-17T14:55:59Z", "updated": "2026-08-17T14:55:59Z", "arxiv_id": "http://arxiv.org/abs/2608.16666v1", "url": "https://arxiv.org/abs/2608.16666v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["ying"], "key_results": ["This paper presents Chronocooked, a reinforcement learning (RL) benchmark suite for studying implicit interval timing in RL agents."], "abstract": "This paper presents Chronocooked, a reinforcement learning (RL) benchmark suite for studying implicit interval timing in RL agents. Inspired by Overcooked, the suite comprises cooking scenarios that require temporal decision making. The tasks and reward functions are designed such that temporal information is unobserved yet critical for optimal performance. The environment is intentionally kept simple to enable controlled experiments and support biologically plausible models. Evaluation metrics are designed to expose limitations in timing abilities of RL agents, and we report baselines using a non-recurrent, a recurrent, and a biologically plausible model. This work ultimately aims to underscore the need to incorporate time perception and temporal processing in artificial agents designed for human robot interaction and deployment in time dependent human societies.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:32:03.905232+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 44986, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Multi-Agent", "Memory / RAG", "Deep Research", "Video Generation", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["phism", "Philip", "ying", "yield"], "matched_methods": ["RLVR / GRPO"], "checked_on": "2026-08-19T03:31:16.021924+00:00"}}
{"id": "evg-58144f90f83cc497", "title": "Reconstruction: A Blind Benchmark for Recovering Research Ideas from Pre-Publication Bibliographies", "authors": ["Shaolong Chen", "Yanlin Fei", "Nazhou Liu", "Xinmiao Yu", "Lei Li", "Rahul Thapa", "Madalina Ciobanu", "Qingqing Mao", "Ritankar Das"], "year": 2026, "published": "2026-08-17T14:44:30Z", "updated": "2026-08-17T14:44:30Z", "arxiv_id": "http://arxiv.org/abs/2608.16645v1", "url": "https://arxiv.org/abs/2608.16645v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.MA"], "pillar": "RL / Alignment / Safety", "methods": ["Multi-Agent", "Deep Research"], "benchmarks": [], "models": ["phies.", "phies"], "key_results": ["We introduce Reconstruction, a blind idea-recovery benchmark that withholds the seed paper and all contemporaneous or future literature, and asks models to propose hypotheses that an independent large language model judge matches against the held-out ground-truth idea.", "Across six scientific domains and 643 evaluated papers, seven frontier models achieve only modest Match rates (approx."], "abstract": "Can a language model recover the true research idea of a published paper when given only that paper's pre-publication bibliography? We introduce Reconstruction, a blind idea-recovery benchmark that withholds the seed paper and all contemporaneous or future literature, and asks models to propose hypotheses that an independent large language model judge matches against the held-out ground-truth idea. A strict anti-leakage protocol-temporal citation cutoff, anonymous reference IDs, and frozen per-paper bibliographies, which prevents prompt-time leakage of the seed idea. Across six scientific domains and 643 evaluated papers, seven frontier models achieve only modest Match rates (approx. 3-15%). We then evaluate a reference-only multi-agent (top 4) pipeline that combines cross-model review with a Swiss tournament over aligned hypothesis slots, without external web search. Cross-model review plus tournament selection raises Match rates to approx. 23-42% across all six domains, which is an observed approx. 2.4x lift over the best single-model baseline. This draft reports the protocol, anti-leakage design, and current results as an arXiv timestamp.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:32:03.905232+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 11593, "fulltext_methods": ["Preference Optimization", "Multi-Agent", "Deep Research"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-5.6-Sol-Pro", "Claude-Opus-4.8", "Claude", "phies", "GLM", "GLM-5.2.", "yield", "ying", "Kimi", "Kimi-K3"], "matched_methods": ["Deep Research", "Multi-Agent"], "checked_on": "2026-08-19T03:32:37.395740+00:00"}}
{"id": "evg-55659e47085e1354", "title": "When Do Explanations Help In-Context Learning? A Comparative Study of Natural Language Explanation Types and Faithfulness", "authors": ["Mahdi Dhaini", "Adam Dejl", "Juraj Vladika", "Volkan Özer", "Barbara Plank", "Gjergji Kasneci"], "year": 2026, "published": "2026-08-17T14:30:03Z", "updated": "2026-08-17T14:30:03Z", "arxiv_id": "http://arxiv.org/abs/2608.16627v1", "url": "https://arxiv.org/abs/2608.16627v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": ["MATH"], "models": ["ying", "yields"], "key_results": ["Our extensive evaluation shows that, on classification-style benchmarks, adding NLEs to few-shot prompts often improves accuracy over few-shot prompting without explanations; among NLE sources, externally generated LLM-NLEs often provide strong downstream utility and remain competitive with human rationales where both are available, whereas self-NLEs are more sensitive to the selection strategy.", "We further show that faithfulness-based selection of self-NLEs yields small average gains overall, but can improve or reduce performance depending on the metric, task, and model.", "Robustness tests with randomly swapped and out-of-distribution rationales indicate partial robustness, suggesting that semantic alignment contributes to performance gains."], "abstract": "Natural language explanations (NLEs) are increasingly used as inputs, for example, as few-shot rationales that influence model behavior in in-context learning (ICL). However, it remains unclear how different types of NLEs compare in their effects on downstream model performance in explanation-augmented prompting. Therefore, we provide a comparative evaluation across six benchmarks and four instruction-tuned models, studying how NLE source (human-written when available, self-generated explanations, generated by an external LLM) and NLE selection (random vs faithfulness-based filtering) affect downstream utility of NLEs when used in ICL settings. Our extensive evaluation shows that, on classification-style benchmarks, adding NLEs to few-shot prompts often improves accuracy over few-shot prompting without explanations; among NLE sources, externally generated LLM-NLEs often provide strong downstream utility and remain competitive with human rationales where both are available, whereas self-NLEs are more sensitive to the selection strategy. On math reasoning, the effects are more model- and source-dependent. We further show that faithfulness-based selection of self-NLEs yields small average gains overall, but can improve or reduce performance depending on the metric, task, and model. Different faithfulness metrics can disagree substantially, affecting which self-NLE examples are selected and their downstream predictive utility. Robustness tests with randomly swapped and out-of-distribution rationales indicate partial robustness, suggesting that semantic alignment contributes to performance gains. Overall, our results provide insights for selecting and reporting explanations that influence model behavior in practical prompting pipelines.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:32:03.905232+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 126092, "fulltext_methods": ["Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Deep Research", "Quant / Trading"], "fulltext_benchmarks": ["GSM8K", "MATH", "BIG-Bench"], "fulltext_models": ["GPT-4o-mini", "Gpt-4o", "o3-mini.", "o3-mini", "o3", "o3.", "Llama-3.1-8B", "Llama-3.3-70B", "Llama-8B", "Llama-70B"], "matched_methods": ["Preference Optimization"], "checked_on": "2026-08-19T03:32:41.212641+00:00"}}
{"id": "evg-56994fb25b3a9448", "title": "Degradation-Aligned Self-Supervised Learning for State of Health Estimation of Lithium-Ion Batteries under Label Sparsity", "authors": ["Jiaqi Yao", "Julia Kowal"], "year": 2026, "published": "2026-08-17T14:14:54Z", "updated": "2026-08-17T14:14:54Z", "arxiv_id": "http://arxiv.org/abs/2608.16612v1", "url": "https://arxiv.org/abs/2608.16612v1", "primary_category": "eess.SP", "categories": ["eess.SP", "cs.AI", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Test results showcase that the proposed ranking-based SSL approach proves to endow the pretrained model with degradation-aligned information from unlabeled data, and after fine-tuning the model can carry out accurate, robust SOH estimation, even when only an extremely limited amount of 1% of unevenly distributed labeled training data is available, where the MAE of 1.718% and RMSE of 2.329% can be achieved on the test cell."], "abstract": "An accurate estimation of the state of health (SOH) underpins a safe and optimized use of the battery system. Although compelling, data-driven SOH estimation models typically require large amounts of high-quality labeled cycling data, while in practice such labels are often sparse in both quantity and coverage. Therefore, in this work, we propose a degradation-aligned self-supervised learning (SSL) framework based on a convolutional neural network-gated recurrent unit (CNN-GRU) model, which learns aging-consistent representations from unlabeled data through a cycle-order ranking objective as the pretext task for pretraining, thereby enabling robust SOH estimation after fine-tuning on sparsely labeled data. Test results showcase that the proposed ranking-based SSL approach proves to endow the pretrained model with degradation-aligned information from unlabeled data, and after fine-tuning the model can carry out accurate, robust SOH estimation, even when only an extremely limited amount of 1% of unevenly distributed labeled training data is available, where the MAE of 1.718% and RMSE of 2.329% can be achieved on the test cell. In addition, in-depth analyses are presented regarding the influences of label distribution of battery degradation data. We believe this work could shed new light on SOH estimation of lithium-ion batteries under label sparsity in real-world applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:32:03.905232+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 81576, "fulltext_methods": ["Preference Optimization", "Interpretability", "Safety / Jailbreak", "Memory / RAG"], "fulltext_benchmarks": [], "fulltext_models": ["phite", "phite-LiFePO4", "ying", "yields", "yielding", "yield", "Yi", "bert", "BERT"], "matched_methods": [], "checked_on": "2026-08-19T03:32:48.127991+00:00"}}
{"id": "evg-c92741f80bb56a99", "title": "BabelSteering: Multilingual Safety Alignment via English Steering Vectors", "authors": ["Emma V. Stein", "Dominik Meier", "Terry Ruas", "Jan Philip Wahle", "Bela Gipp"], "year": 2026, "published": "2026-08-17T13:40:16Z", "updated": "2026-08-17T13:40:16Z", "arxiv_id": "http://arxiv.org/abs/2608.16577v1", "url": "https://arxiv.org/abs/2608.16577v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Safety / Jailbreak"], "benchmarks": ["MMLU"], "models": ["ying", "Gemma 7B"], "key_results": ["In this work, we investigate whether safety signals learned from a high-resource language, like English, can improve multilingual safety."], "abstract": "Large language models (LLMs) are deployed globally in high-stakes settings, yet most safety research and alignment efforts remain concentrated on English. Thus, users interacting with LLMs in other languages may encounter weaker safeguards despite relying on the same systems for similarly sensitive tasks. In this work, we investigate whether safety signals learned from a high-resource language, like English, can improve multilingual safety. We propose BabelSteering, an activation steering method that acts as a lightweight inference- time intervention, using refusal directions derived from English safety supervision to generalize across languages. Our evaluation includes eight languages and jointly measures refusal of harmful requests, over-refusal, and general task utility. The results show that BabelSteering increases the refusal of harmful requests across languages, with only a marginal to no reduction in task utility but with some increase in refusal of pseudo-harmful prompts. For example, for Gemma 7B, we see an average increase in the refusal of harmful prompts across languages of 11 percentage points (pp), with individual languages like Bengali seeing an increase of 17 pp, with no loss of utility on Global MMLU, while pseudo-harmful refusals increase by 13 pp on average. We also introduce a multilingual translation-and-evaluation pipeline to facilitate future work on cross-lingual safety interventions. Overall, our findings suggest that activation steering may provide a practical, low- cost mechanism for extending English-derived safety signals to other languages. Warning: this paper contains examples with unsafe content", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:32:03.905232+00:00"}
{"id": "evg-6742b060cb568d54", "title": "Le Critique: Privileged Value Functions for LLM Reinforcement Learning", "authors": ["Siddarth Venkatraman", "Matthieu Dinot", "Laurence Aitchison"], "year": 2026, "published": "2026-08-17T15:49:36Z", "updated": "2026-08-17T15:49:36Z", "arxiv_id": "http://arxiv.org/abs/2608.16739v1", "url": "https://arxiv.org/abs/2608.16739v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["We propose two complementary strategies to improve the performance of value function RL: 1) Privileged Value Functions (PVF) which provide an elegant mechanism to inject additional task-relevant token-level signal without biasing the policy objective; 2) TETHER, a baseline that adaptively interpolates between group-relative and value baselines depending on the value function accuracy.", "Across several reasoning tasks, both strategies consistently improve over the standard value function baseline, and are competitive with or outperform mean-baseline GRPO."], "abstract": "Reinforcement learning algorithms for Large Language Models (LLMs) are largely distinguished by their variance reduction strategy. Group-relative methods like GRPO reduce gradient variance by sampling multiple rollouts per prompt, but provide only sequence-level credit. Training is also blocked by straggler rollouts, reducing throughput and increasing off-policyness. Learned value functions theoretically address both problems, providing token-level advantages without requiring large groups. However, additional infrastructure engineering challenges combined with the practical success of critic-free methods have made it difficult to justify their inclusion in RL pipelines. We propose two complementary strategies to improve the performance of value function RL: 1) Privileged Value Functions (PVF) which provide an elegant mechanism to inject additional task-relevant token-level signal without biasing the policy objective; 2) TETHER, a baseline that adaptively interpolates between group-relative and value baselines depending on the value function accuracy. Across several reasoning tasks, both strategies consistently improve over the standard value function baseline, and are competitive with or outperform mean-baseline GRPO.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 62532, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Distillation", "Preference Optimization", "Multi-Agent", "VLM", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["Llama 2", "Qwen3-4B-Instruct-2507", "Qwen3.5-4B", "Qwen", "Qwen3", "Qwen3.5", "DeepSeek-R1", "DeepSeek-AI", "DeepSeekMath", "R1"], "matched_methods": ["RLVR / GRPO"], "checked_on": "2026-08-18T07:48:23.170562+00:00"}}
{"id": "evg-6c5aaefababff9fc", "title": "KV-Pipe: On the Relation Between KV Sharing and Pipeline Parallel Efficiency in LLMs", "authors": ["Maryam Dialameh", "Hossein Rajabzadeh", "Harish Krishnamoorthy Murali", "Walid Ahmed", "Weiwei Zhang", "Hyock Ju Kwon"], "year": 2026, "published": "2026-08-16T22:01:16Z", "updated": "2026-08-16T22:01:16Z", "arxiv_id": "http://arxiv.org/abs/2608.15943v1", "url": "https://arxiv.org/abs/2608.15943v1", "primary_category": "cs.DC", "categories": ["cs.DC"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache", "Long Context"], "benchmarks": [], "models": [], "key_results": ["Across multiple pipeline-parallel configurations, KV-Pipe consistently improves utilization and throughput, achieving up to \\textbf{9.2\\%} higher training MFU and up to a \\textbf{9.8\\%} reduction in iteration time, with larger gains at higher pipeline-parallel degrees where stage imbalance is amplified."], "abstract": "Pipeline parallelism (PP) is widely used to scale large language model (LLM) training, but its efficiency is often limited by stage imbalance and pipeline bubbles. Meanwhile, cross-layer KV sharing has primarily been studied as a mechanism for reducing KV-cache costs during inference, without examining how KV reuse reshapes pipeline workloads. We present \\textbf{KV-Pipe}, a stage-aware KV-sharing mechanism that turns KV reuse into a pipeline-balancing control knob. KV-Pipe starts from the tail stage, converts selected attention layers to cross-layer KV sharing in a tail-first order, and iteratively retargets the current bottleneck to drive the FLOPs Imbalance Ratio (FIR) toward $1$. The procedure is performed offline and requires only a pipeline partition and per-layer FLOPs estimates, introducing negligible runtime overhead and requiring no online tuning. Across multiple pipeline-parallel configurations, KV-Pipe consistently improves utilization and throughput, achieving up to \\textbf{9.2\\%} higher training MFU and up to a \\textbf{9.8\\%} reduction in iteration time, with larger gains at higher pipeline-parallel degrees where stage imbalance is amplified. Furthermore, the same KV-sharing mechanism provides an inference-side benefit by reducing KV-cache growth and redundant KV projection work, resulting in higher decoding throughput for long-context workloads. These results identify KV layout as a system--architecture degree of freedom for jointly improving pipeline-parallel training efficiency and long-context inference.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-2e6a4a8e28a606d4", "title": "Aborted but Not Forgotten: KV-Cache Retention Breaks Rollback Consistency in Language Agents", "authors": ["Guijia Zhang", "Harry Yang"], "year": 2026, "published": "2026-08-16T21:49:38Z", "updated": "2026-08-16T21:49:38Z", "arxiv_id": "http://arxiv.org/abs/2608.15939v1", "url": "https://arxiv.org/abs/2608.15939v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "KV Cache"], "benchmarks": [], "models": ["ying"], "key_results": ["Stateful language agents assume a rejected branch can be taken back by clearing it from the application transcript."], "abstract": "Stateful language agents assume a rejected branch can be taken back by clearing it from the application transcript. We show this breaks when the serving session retains key/value (KV) state across the logical abort: the model can continue attending to content the application believes it discarded. We formalize the missing guarantee as rollback consistency: a complete abort must restore the state the model attends, not just the transcript. The key failure is cross-layer: a correct logical rollback need not compose with retained inference state, and the gap can remain invisible to the application. To isolate cache effects from text effects, we introduce a same-token/different-cache audit that holds decision-step tokens identical while varying only whether the cached prefix is stale or rebuilt from committed state. Across seven open-weight families (3.8B-36B), retained KV alone flips a typed protected effect in 25 of 63 audited cells, while attacker tokens are absent from the served request in all 63; rebuilding the cache closes every cell. The channel reproduces in an end-to-end session application, on the default Hugging Face Transformers cache-reuse path, and under LangGraph time-travel, where verified logical rollback can still leave attended KV stale. Susceptibility varies across models, but the underlying attended-state integrity violation is structural. We rule out position and length confounds, generalize across protected effects, policy structures, and a cache-isolated Mixture-of-Experts model, and show that transaction-local cache restoration closes the channel without requiring a global cache flush. All headline results are deterministic and reproducible from released artifacts.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-e9cf1d0c275420ed", "title": "Cross-Entropy Risk Estimation for Language Models: Inconsistency Must Be Dense, and the Holdout Method Is No Exception", "authors": ["Hanti Lin"], "year": 2026, "published": "2026-08-16T15:24:52Z", "updated": "2026-08-16T15:24:52Z", "arxiv_id": "http://arxiv.org/abs/2608.15798v1", "url": "https://arxiv.org/abs/2608.15798v1", "primary_category": "cs.LG", "categories": ["cs.LG", "stat.ML"], "pillar": "Efficient Training & Inference", "methods": ["Long Context"], "benchmarks": [], "models": ["ying"], "key_results": ["Language models are compared by their held-out per-token cross-entropy risk---the quantity scaling laws are fitted to."], "abstract": "Language models are compared by their held-out per-token cross-entropy risk---the quantity scaling laws are fitted to. We show that it cannot be consistently estimated. Consistency, or convergence to the estimand, is defined relative to a \\emph{possible state of the world}: a pair consisting of a data-generating distribution and a model we turn out to train. Quantifying over models as well as data-generating mechanisms is essential, because what decides whether a model's risk is estimable is a tail property of the distribution its weights induce, which no sample reveals. The per-token cross-entropy risk is hard to estimate because of a topological fact: among the possible states, finite risk and infinite risk each lie arbitrarily close to every instance of the other. Consequently no estimator---not merely the holdout average---is consistent at every state at which the risk is defined. Worse, inconsistent estimation persists under both bounding the expected sequence length and restricting to full-support models; and in that restricted setting the states at which inconsistency occurs are even dense. Two interesting ways out are identified, and neither is free. Way out 1: using a bounded context window, we can floor a model's next-token probabilities, making its risk finite exactly when the data-generating distribution has finite expected sequence length---a new, statistical rationale for a choice that was made on computational grounds, though the assumption it substitutes is itself beyond the reach of any test. Way out 2: reporting the risk only when it falls below a threshold fixed in advance restores consistency, at no cost to what model selection actually requires---but we need to recognize that the goal of estimation is revised.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-e2b424c8ce50cdaa", "title": "KV-Rescue: Recovering Reasoning Language Model KV Eviction Loss via Stepwise Interleaving", "authors": ["Minsoo Cheong", "Woosang Lim", "Vincent-Daniel Yun", "Sungjoo Yoo"], "year": 2026, "published": "2026-08-16T15:23:25Z", "updated": "2026-08-16T15:23:25Z", "arxiv_id": "http://arxiv.org/abs/2608.15797v1", "url": "https://arxiv.org/abs/2608.15797v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Chain-of-Thought", "KV Cache", "Memory / RAG"], "benchmarks": ["MATH"], "models": ["Qwen2.5-Math"], "key_results": ["KV-cache eviction caps the memory cost of long reasoning traces but is inherently lossy because the model decodes from a partial view of its history."], "abstract": "KV-cache eviction caps the memory cost of long reasoning traces but is inherently lossy because the model decodes from a partial view of its history. Under aggressive budgets, this not only lowers accuracy but can also cause runaway degeneration, where the model produces incoherent or repetitive tokens until reaching the length limit. We characterize much of this loss as an information gapf caused by missing context, rather than a capability gap caused by limited model capacity. An evicted 7B model and a full-context 1.5B model make complementary errors, and an oracle choice between their answers recovers 79% of the accuracy gap to the full-KV 7B model. Based on this observation, we propose KV-Rescue, a training-free inference framework that bridges the information gap introduced by KV eviction using a lightweight full-context helper. KV-Rescue interleaves reasoning steps from the two models into a shared trajectory. An online detector uses entropy and compressibility to terminate the generation of incoherent or repetitive base-model candidates early. Across five math benchmarks with Qwen2.5-Math 7B and 72B, KV-Rescue recovers an average of 87% of the accuracy lost to eviction at eviction budget B=64. A decode-cost analysis further shows that preventing runaway degeneration cuts base-model token generation by 43% on average.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-2d4a155b21ee6ae9", "title": "Routing Divergence Is Not Evidence of Behavioral Influence in Same-Weight MoE Self-Distillation", "authors": ["Cedric Caruzzo", "Donggeun Yoo", "Tae Soo Kim"], "year": 2026, "published": "2026-08-16T15:07:17Z", "updated": "2026-08-16T15:07:17Z", "arxiv_id": "http://arxiv.org/abs/2608.15787v1", "url": "https://arxiv.org/abs/2608.15787v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "Distillation"], "benchmarks": [], "models": [], "key_results": ["Two Mixture-of-Experts (MoE) forward passes can share every weight yet route the same token through different experts."], "abstract": "Two Mixture-of-Experts (MoE) forward passes can share every weight yet route the same token through different experts. This creates a possible blind spot in same-weight self-distillation, where a demonstration-conditioned teacher supervises a query-only student. We study this mismatch in its single-step form, with frozen weights rather than as a proxy for a full training trajectory. An exact blockwise decomposition separates a routing term, which changes gates at fixed content, from a dense-like content term. Across seven open-weight checkpoints and two domains, the routing term spans only $1.6\\times$ as a fraction of block output, while its residual-stream exposure spans $3.2\\times$. Exposure is ordered by the routed block's share of the residual. Scaling the always-on backbone in two confirmatory models moves exposure monotonically; common-mode controls support a mass-and-coherence mechanism rather than denominator dilution alone. Preregistered PubMedQA patches on three models show that the full routing term moves outputs by less than half the natural context effect and is largely reproduced by matched-norm noise, whereas the content term is strongly direction-specific. Scale and merged-expert probes show that the narrow block-level range is not universal, although exposure remains small at the tested boundaries. Router movement alone is therefore not evidence of behavioral influence: measure exposure first, and use a behavioral intervention when the decision matters.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-f1b72bb3385220b7", "title": "TaoLive Digital Avatar Agent Technical Report: Training Agents to Evolve with Their Harness", "authors": ["TaoLive AIGC LLM Team", "Yuhan Sun", "Wenhao Lin", "Yongdong Luo", "Yibo Hu", "Meiguang Jin", "Junfeng Ma", "Weihang Pan", "Jiaxin Zhao", "Zulong Chen"], "year": 2026, "published": "2026-08-16T14:32:56Z", "updated": "2026-08-16T14:32:56Z", "arxiv_id": "http://arxiv.org/abs/2608.15763v1", "url": "https://arxiv.org/abs/2608.15763v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "Distillation", "Preference Optimization", "Multi-Agent"], "benchmarks": ["IFEval"], "models": [], "key_results": ["By contrast, fixed-Harness SFT reduces IFEval by 7.7 points.", "In a controlled complete-agent replay on one NVIDIA H20 GPU with MTP enabled, the system achieves 3.407 s P50 and 8.114 s P95 latency."], "abstract": "AI-powered digital-avatar streamers in live e-commerce must answer product questions, engage viewers, and execute changing business strategies in real time. This requires low latency, factual and effective replies, and rapid adaptation to updated campaign, compliance, and style requirements. We develop an evolvable Harness that decouples Skills, Hooks, system prompts, and tools from model weights, allowing runtime behavior to change without retraining. However, Harness evolution creates a moving execution environment: compact models fine-tuned on one configuration may memorize names, schemas, and prompt templates rather than follow the Harness currently provided, while stronger zero-shot models are too slow for real-time use. We address this tension with Harness-Aware Training (HAT), which makes Harness states part of the training distribution. HAT applies task-preserving Harness-State Augmentation (HSA) to Skills, tool schemas, prompt structures, and interaction constraints, and comprises three stages: HSA-based supervised fine-tuning, general on-policy distillation to recover general capabilities, and HSA-based agentic reinforcement learning in a production-informed live-room simulator. Across four evaluation sets with more than 4,500 cases, our compact 35B model scores 94.8 on real-world Live-Stream QA, versus 80.3 for the base model and 93.0 for the strongest evaluated general LLM, while scoring 94.6 on Harness-Variant QA and retaining 83.5 on IFEval. By contrast, fixed-Harness SFT reduces IFEval by 7.7 points. In a controlled complete-agent replay on one NVIDIA H20 GPU with MTP enabled, the system achieves 3.407 s P50 and 8.114 s P95 latency. These results show that HAT produces a latency-feasible compact agent that remains effective under evaluated Harness changes without sacrificing general instruction following.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-b80d063eac10a100", "title": "Large Models for Small Devices: Recent Advances and Empirical Analysis of Edge AI Deployment", "authors": ["Subhransu Das", "Jiaming Cheng", "Arnav Kumar", "Sadia Afrose", "Mingzhe Han", "Michael Silagy", "Shreya Palande", "Brijesh Soni", "Rajiv Ramnath"], "year": 2026, "published": "2026-08-16T11:52:13Z", "updated": "2026-08-16T11:52:13Z", "arxiv_id": "http://arxiv.org/abs/2608.15693v1", "url": "https://arxiv.org/abs/2608.15693v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Preference Optimization"], "benchmarks": ["SQuAD"], "models": ["Qwen3.5"], "key_results": ["Running large AI models on resource-constrained edge devices requires model compression to reduce model size and computation."], "abstract": "Running large AI models on resource-constrained edge devices requires model compression to reduce model size and computation. What compresses well, however, need not deploy well. We survey dozens of recent works that report compression results on real hardware and extract practical deployment guidelines from them. Following these guidelines, we deploy compact language and image models on GPU, CPU, and Raspberry Pi platforms across question answering and image segmentation. No single technique wins across tasks. For question answering, Qwen3.5 0.8B reaches 93.85 SQuAD F1 and 92 EM under Q5_K_M GGUF quantization, while structured pruning at the same precision costs 16 F1 at a 1% ratio. For segmentation, the ranking reverses: default quantization leaves parameters and MACs unchanged, whereas pruning cuts model size by nearly 80% at near-constant mIoU. Pruning can even inflate the deployed artifact by 21-49% by breaking k-quant super-block alignment; combined with longer, less format-compliant outputs, this raises Raspberry Pi latency up to 3.4x. Compression can also manufacture the appearance of competence rather than destroy it visibly: one LoRA-recovered variant stays fully parseable and holds 71% strict BoolQ accuracy while sending 97 of 100 predictions to a single class, at 52.6% balanced accuracy. We explain these effects through neural-flow graph analysis and prefill-decode-level latency decomposition, and condense them into task-specific deployment research directions. The right technique depends on the task, the model, and the hardware. Our experiment code and artifacts are open-sourced at https://github.com/Arnavvvkumar/deployment", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-85ef206d2a6127e2", "title": "Adaptive Heterogeneous Compression for Resource-Efficient Federated Knowledge Distillation", "authors": ["Chenwang Liu", "Yijun Liu", "Chang Liu", "Xu Zhang", "Pengchao Han"], "year": 2026, "published": "2026-08-16T10:08:26Z", "updated": "2026-08-16T10:08:26Z", "arxiv_id": "http://arxiv.org/abs/2608.15660v1", "url": "https://arxiv.org/abs/2608.15660v1", "primary_category": "cs.DC", "categories": ["cs.DC", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["An efficiency-aware reward is designed by jointly considering local optimization improvement, global knowledge alignment, and execution time."], "abstract": "Federated learning (FL) enables privacy-preserving distributed model training but faces challenges from heterogeneous model architectures and limited communication resources at the network edge. Federated knowledge distillation (FedKD) alleviates model heterogeneity by combining prototype-wise parameter aggregation and knowledge transfer across heterogeneous models. However, transmitting gradients still introduces considerable communication overhead, while existing compression approaches typically apply a uniform strategy across clients and ignore their diverse model characteristics and resource capacities. To address this issue, we propose a heterogeneous compression framework for FedKD that enables each client to select a compression strategy from a candidate strategy set. We formulate the compression strategy selection problem as a non-stationary stochastic multi-armed bandit (MAB), where each arm corresponds to a compression strategy. An efficiency-aware reward is designed by jointly considering local optimization improvement, global knowledge alignment, and execution time. Based on this formulation, we develop an Adaptive heterogeneouS Compression algorithm for fEderated kNowledge Distillation (ASCEND), which employs an exponential moving average (EMA)-enhanced $ε$-greedy policy to balance exploration and exploitation. Experimental results on multiple datasets demonstrate that ASCEND effectively adapts to heterogeneous model and resource settings, reducing communication overhead and training time while maintaining competitive model accuracy.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-d409d78d17769114", "title": "FluxBin: Flexible LUT-based Ultra-low-bit LLM Inference by Algorithm-Kernel Synergy", "authors": ["Qingyao Yang", "Runming Yang", "He Xiao", "Wendong Xu", "Junyu Chen", "Haobo Liu", "Chenchen Ding", "Ruihan Hu", "Yik-Chung Wu", "Ngai Wong"], "year": 2026, "published": "2026-08-16T08:01:06Z", "updated": "2026-08-16T08:01:06Z", "arxiv_id": "http://arxiv.org/abs/2608.15602v1", "url": "https://arxiv.org/abs/2608.15602v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Memory / RAG"], "benchmarks": [], "models": ["FluxBin", "FluxBin achieves up to", "FluxBin. cs.LG cs.AI"], "key_results": ["Extensive evaluations demonstrate FluxBin achieves up to $5.92\\times$ speedup and $10.19\\times$ energy savings across diverse model architectures, delivering comparable accuracy to heavily fine-tuned methods."], "abstract": "While binary quantization theoretically promises extreme compression and acceleration for Large Language Models (LLMs), existing research often overlooks the necessity of specialized hardware kernels, thus failing to unleash the full acceleration potential due to persistent reliance on expensive floating-point arithmetic or runtime dequantization overheads. To bridge this gap, we propose FluxBin (\\textbf{F}lexible \\textbf{L}UT-based \\textbf{U}ltra-low-bit e\\textbf{X}ecution with \\textbf{Bin}ary bases), an algorithm-kernel co-design that synergizes post-training quantization with a highly optimized CUDA kernel. Algorithmically, we introduce Decoupled Row-Column Binary Decomposition to enhance representational capacity while maintaining hardware efficiency, complemented by a Hessian-guided saliency-aware hybrid bases that preserve critical information. At the kernel level, we implement a Lookup Table Building Approach with Scale Fusion to reduce floating-point arithmetic, featuring a Virtual Columnar Mapping that transforms irregular, sparse, and salient matrices into dense execution. Extensive evaluations demonstrate FluxBin achieves up to $5.92\\times$ speedup and $10.19\\times$ energy savings across diverse model architectures, delivering comparable accuracy to heavily fine-tuned methods. This effectively enables the deployment of 70B-scale models on one single A100 GPU with a $4\\times$ memory reduction. Code is available at https://github.com/nicyyyy/FluxBin.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-4f3424408d7b4e9a", "title": "GraniKV: Asymmetric Granularity KV-Cache Paging for Multi-Agent Systems with Long Shared Prefix", "authors": ["Jinhyun Jeon", "Sungjoo Yoo"], "year": 2026, "published": "2026-08-16T07:19:49Z", "updated": "2026-08-16T07:19:49Z", "arxiv_id": "http://arxiv.org/abs/2608.15584v1", "url": "https://arxiv.org/abs/2608.15584v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache", "Multi-Agent", "Memory / RAG"], "benchmarks": [], "models": ["Llama-3.1-8B", "Qwen-2.5-14B", "Qwen-2.5-32B"], "key_results": ["Production paged-serving engines apply uniform paging granularity to the KV cache, even though the two regions of a multi-agent workload have opposite storage requirements: a long shared prefix demands contiguity, while the per-request suffix demands fine-grained allocation."], "abstract": "Production paged-serving engines apply uniform paging granularity to the KV cache, even though the two regions of a multi-agent workload have opposite storage requirements: a long shared prefix demands contiguity, while the per-request suffix demands fine-grained allocation.\n  We present \\textbf{GraniKV}, a KV-cache layer that allocates the shared prefix in a contiguous HOT pool and the suffix in a token-level COLD pool, combined with a per-step dispatcher which selects the appropriate backend among dual backends for each regime (compute-, memory-, or communication-bound). To the best of our knowledge, GraniKV is the first system to apply asymmetric paging granularity to the KV cache of a production paged-serving engine.\n  At $L_p{=}16$\\,K shared tokens GraniKV reaches $\\mathbf{2.16\\times}$, $\\mathbf{1.98\\times}$, and $\\mathbf{1.57\\times}$ output-token throughput over the production baseline on Llama-3.1-8B/TP=1, Qwen-2.5-14B/TP=2, and Qwen-2.5-32B/TP=4. The gain decomposes: cascade attention integration contributes the majority at saturation; the asymmetric storage layer adds $1.05$--$1.15\\times$ end-to-end while being what makes the batched-GEMM prefix backend possible at all. Under heterogeneous multi-agent serving with \\emph{distinct} prompts of different lengths, the attribution inverts: GraniKV sustains $\\mathbf{1.95\\times}$ while batch-global cascade collapses to parity --- the storage layer alone carries the win in the regime that motivates the paper.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-bd8ec5c6350ef3db", "title": "Learning Generalizable Reconstruction of High-Dimensional Neural Dynamics", "authors": ["Anima Kujur", "Zahra Monfared"], "year": 2026, "published": "2026-08-17T13:35:38Z", "updated": "2026-08-17T13:35:38Z", "arxiv_id": "http://arxiv.org/abs/2608.16569v1", "url": "https://arxiv.org/abs/2608.16569v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["yielded"], "key_results": ["On 200,000-sample hippocampal recordings, PCA-DMD outperformed Classical DMD, SpDMD, MrDMD, and HODMD, achieving KLD=0.0761 and HD=0.0847."], "abstract": "Accurate reconstruction of long-duration neural recordings is challenging because local field potentials (LFPs) are high-resolution, multichannel, transient, and variable across subjects. We present PCA-DMD, a scalable operator-theoretic framework that segments LFP recordings into overlapping windows, projects them into a compact PCA space, learns linear Koopman evolution in the latent space, and reconstructs continuous signals through inverse projection and overlap-add aggregation. On 200,000-sample hippocampal recordings, PCA-DMD outperformed Classical DMD, SpDMD, MrDMD, and HODMD, achieving KLD=0.0761 and HD=0.0847. In all-pair cross-subject zero-shot generalization at 300,000 samples, correlations were 0.9504-0.9800, with HD=0.0010-0.0072 and KLD=0.0005-0.0022, without target-subject fine-tuning. Out-of-sample temporal prediction showed close one-step agreement on temporally held-out LFP segments across the unseen interval and multiple channels. Scalability analysis from 400,000 to 900,000 samples showed stable zero-shot reconstruction, with mean correlation remaining about 0.965-0.968 while computational cost increased predictably. External validation on an independent 93-channel Allen Neuropixels recording yielded mean and median channel-wise correlations of 0.7427 and 0.7990, respectively. Koopman spectral and mode analyses revealed dominant eigenvalues concentrated near the unit circle. PCA-DMD therefore provides an interpretable, generalizable, and computationally scalable framework for reconstructing high-dimensional neural dynamics.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-bb0e7f0f31686da1", "title": "STAGE: Controlled Objective Admission for Multi-Preference LLM Alignment", "authors": ["Yongqi Tong", "Zhenyu Zhang", "Ruirui Wang", "Kewei Fu", "Shaoqing Lin", "Sijie Dong", "Jiang-Ming Yang", "Xin Zhang", "Jianshe Li"], "year": 2026, "published": "2026-08-17T13:24:11Z", "updated": "2026-08-17T13:24:11Z", "arxiv_id": "http://arxiv.org/abs/2608.16553v1", "url": "https://arxiv.org/abs/2608.16553v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization", "Interpretability"], "benchmarks": [], "models": [], "key_results": ["Multi-preference alignment is often framed as scalarization: combine reward dimensions, then optimize."], "abstract": "Multi-preference alignment is often framed as scalarization: combine reward dimensions, then optimize. This leaves a temporal decision underspecified: when should each preference dimension enter policy optimization? We propose \\methodname, a stability-guided active-set controller for controlled objective admission. \\methodname starts from a small active set, retains admitted objectives, and expands when reward-deviation gates indicate low recent deviation or a patience budget is exhausted. A probing phase estimates a hard-to-easy order, and adaptive weighting emphasizes underperforming active dimensions. Automatic evaluations with 15 training preferences and 16 held-out benchmark columns show that \\methodname obtains higher averages than simultaneous scalarization and shared-budget adapted baselines. Component ablations and expansion dynamics further support cumulative retention, gated admission, and probing-derived ordering as useful design choices in this setting. These results position objective-entry timing as a concrete control variable in reward-vector RLHF.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-b6c1942e38d115cf", "title": "Listen, Reason, and Segment: Aligning LALMs with Editorial Judgment for Media Chapterization", "authors": ["Tony Alex", "Wish Suharitdamrong", "Sara Atito", "Armin Mustafa", "Muhammad Awais", "Philip J. B. Jackson", "Jiankang Deng", "Ismail Elezi"], "year": 2026, "published": "2026-08-17T13:14:01Z", "updated": "2026-08-17T13:14:01Z", "arxiv_id": "http://arxiv.org/abs/2608.16539v1", "url": "https://arxiv.org/abs/2608.16539v1", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.AI", "cs.CL", "eess.AS"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Chain-of-Thought", "Preference Optimization"], "benchmarks": [], "models": ["R1-Zero", "R1", "ying"], "key_results": ["Applying GRPO directly without a Supervised Fine-Tuning (SFT) cold start, AudioChaps-R1-Zero already improves average F1 by 33 points over the state-of-the-art LALM Audio-Flamingo-3-Think.", "The AudioChaps framework produces our final aligned LALM, AudioChaps-R1, which improves average F1 by 49 points."], "abstract": "Large Audio Language Models (LALMs) have made rapid progress on standardized benchmarks, yet their deployment in practical media workflows, curation, archival indexing, and content distribution remains largely unrealized. We identify automated audio chapterization, the task of segmenting continuous audio streams into thematically coherent chapters, as a demanding and commercially consequential setting that exposes this gap. Chapterization is challenging because boundaries are defined less by objective acoustic events than by subjective editorial judgment, requiring models to reason sequentially over long acoustic contexts and approximate creator-authored boundary decisions. We present AudioChaps, a post-training framework for aligning end-to-end LALMs for this task via Group Relative Policy Optimization (GRPO) guided by Chain-of-Thought (CoT) reasoning. To support training and evaluation, we curate three datasets: AudioChaps-Alignment, derived from creator-annotated chapter boundaries on YouTube; AudioChaps-CoT, which provides structured supervision for well-formatted, high-quality, and evidence-grounded boundary reasoning; and AudioChaps-Eval, a held-out benchmark for audio chapterization. Applying GRPO directly without a Supervised Fine-Tuning (SFT) cold start, AudioChaps-R1-Zero already improves average F1 by 33 points over the state-of-the-art LALM Audio-Flamingo-3-Think. The AudioChaps framework produces our final aligned LALM, AudioChaps-R1, which improves average F1 by 49 points. These results demonstrate that GRPO-trained LALMs can reliably transform unstructured auditory streams into navigable, structured media. Our code, models, and dataset resources will be released upon acceptance at https://github.com/ta012/AudioChaps.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-b6e4975833979435", "title": "LLMs for Zero-Shot Threat Detection via Structured Risk Indicators", "authors": ["Abdullah Alghamdi", "Siamak Layeghy", "Marius Portmann"], "year": 2026, "published": "2026-08-17T12:47:38Z", "updated": "2026-08-17T12:47:38Z", "arxiv_id": "http://arxiv.org/abs/2608.16508v1", "url": "https://arxiv.org/abs/2608.16508v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG", "cs.NI"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["All configurations outperform the previous state-of-the-art LLM-based framework (GABM), with the best configuration improving the F1-score by 11.40 percentage points on CERT r5.2 and 31.50 percentage points on PicoDomain.", "Results further show that retrieval mainly benefits weaker LLMs by generating more discriminative risk indicators, whereas stronger models achieve comparable performance without retrieved context."], "abstract": "We propose a two-stage large language model (LLM) framework for zero-shot detection of insider threats and advanced persistent threats (APTs) from heterogeneous security logs. The framework models user activity as chronological timelines and incorporates retrieval-augmented generation (RAG) to provide personalised behavioural context from each user's historical activity. Rather than performing end-to-end classification directly from raw logs, it first generates structured, interpretable sets of threat-specific risk indicators, which are then classified jointly across temporal sequences to capture attack patterns spanning multiple windows.The framework is evaluated on two benchmark datasets, CERT r5.2 for insider threat detection and PicoDomain for APT detection, using four combinations of two open-weight LLMs under both retrieval and non-retrieval settings. All configurations outperform the previous state-of-the-art LLM-based framework (GABM), with the best configuration improving the F1-score by 11.40 percentage points on CERT r5.2 and 31.50 percentage points on PicoDomain. Results further show that retrieval mainly benefits weaker LLMs by generating more discriminative risk indicators, whereas stronger models achieve comparable performance without retrieved context. The most effective assignment of LLMs to the two stages depends on the dataset. These findings show that the quality of the generated risk indicators is the main driver of zero-shot cyber threat detection performance.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-0f4ba78757467e6c", "title": "Density-Reweighted Entropic Optimal Transport: Decoupling Geometry from Sampling Density", "authors": ["Keyi Li", "Yuval Kluger", "Boris Landa"], "year": 2026, "published": "2026-08-17T12:46:03Z", "updated": "2026-08-17T12:46:03Z", "arxiv_id": "http://arxiv.org/abs/2608.16506v1", "url": "https://arxiv.org/abs/2608.16506v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG", "stat.AP", "stat.ME"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["yielding", "ying"], "key_results": ["Dataset alignment is a central step in data analysis across science and engineering, where the goal is to match observations between datasets."], "abstract": "Dataset alignment is a central step in data analysis across science and engineering, where the goal is to match observations between datasets. Entropic Optimal Transport (EOT) offers a computationally tractable framework for this task by encoding cross-dataset affinities in a transport plan. However, when two datasets are sampled from geometrically similar low-dimensional structures with substantially different sampling densities, the EOT plan may match points by relative sampling density rather than geometric proximity, yielding geometrically misleading correspondences. To address this issue, we propose a density-reweighted EOT framework in which the influence of sampling density on the transport plan can be discounted to a desired degree, ranging from standard EOT to alignment driven purely by underlying geometry. Under suitable regularity conditions, we establish convergence of the reweighted EOT plan to a family of population-level plans whose dependence on sampling density is made explicit. Through simulations, we show that our approach recovers geometrically faithful correspondences, improving over related EOT-based frameworks when datasets exhibit substantial sampling density disparity.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-31b2d4d30768b864", "title": "Vantage: Availability-Graded Broadcast for Signature-Free BFT", "authors": ["Nikita Polyanskii"], "year": 2026, "published": "2026-08-17T12:42:46Z", "updated": "2026-08-17T12:42:46Z", "arxiv_id": "http://arxiv.org/abs/2608.16504v1", "url": "https://arxiv.org/abs/2608.16504v1", "primary_category": "cs.DC", "categories": ["cs.DC", "cs.CR"], "pillar": "RL / Alignment / Safety", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Unresolved tips are sealed later through a signature-free control log, by resolutions each correct party checks against its own recorded responses; a crash-only silent view is skipped by a quorum of skip votes without the log."], "abstract": "Digital signatures make blocks and votes transferable evidence: one party can prove to another what a third party said. Authenticated channels convince only the direct receiver, so existing high-throughput signature-free protocols complete an availability vote or a broadcast instance for each data block before any proposer may order it. We present Vantage, a partially synchronous Byzantine fault-tolerant protocol for $n \\ge 3f+1$ parties that uses only authenticated channels and collision-resistant hashing. Parties publish blocks on hash-linked author lanes; each view's proposer pairs a quorum-available core manifest of lane frontiers with an optimistic tip manifest of freshly received blocks. A new primitive, Availability-Graded Broadcast (AGB), makes the core irrevocable on a quorum of first-hand responses while grading, rather than blocking on, the tip. Unresolved tips are sealed later through a signature-free control log, by resolutions each correct party checks against its own recorded responses; a crash-only silent view is skipped by a quorum of skip votes without the log. AGB makes a published block proposal-eligible one message delay after publication, matching signed optimistic-tip designs. When all parties are correct and message delays are $δ$, a data-only proposal seals within $2δ$ of its send on all $n$ first-hand acknowledgments, so a block is sequenced within $4δ$ of publication, and within $3δ$ when publication aligns with the next proposal. We prove safety under asynchrony; liveness holds after the Global Stabilization Time. On an emulated ten-region WAN with 100 parties, Vantage has the lowest median latency among the nearest signature-based and signature-free protocols at every accepted load and sequences 239k 512-byte transactions per second with median latency below 500 ms.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-49c0de6821dd6940", "title": "Graph Machine Learning: An Opportunity for Power Systems", "authors": ["Martin Sadric", "Sebastian Pütz", "Christian Nauck", "Veit Hagenmeyer", "Frank Hellmann", "Dirk Witthaut", "Benjamin Schäfer"], "year": 2026, "published": "2026-08-17T12:33:46Z", "updated": "2026-08-17T12:33:46Z", "arxiv_id": "http://arxiv.org/abs/2608.16494v1", "url": "https://arxiv.org/abs/2608.16494v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CE", "eess.SY"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Safety / Jailbreak", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["We further derive a structured requirements catalog for ML-ready power grid benchmarks, intended to guide future dataset development and improve reproducibility across studies."], "abstract": "Modern power systems face growing operational complexity driven by the integration of renewable energy sources, decentralization, and the need for real-time decision-making across a wide range of timescales. Addressing these challenges traditionally relies on model-based methods that, while accurate, can be too slow for operational demands. Machine learning (ML) has therefore emerged as a faster, data-driven alternative. As grid topology plays a central role in power system operation, graph machine learning (GML) methods offer a natural framework for incorporating topological dependencies as an inductive bias. We survey nearly 800 papers at the intersection of GML and power systems, covering forecasting, state estimation, optimization, control, fault diagnosis, and cybersecurity. Power systems constitute an unusually rich benchmark setting for GML, as they combine hard physical constraints, multi-scale dynamics, safety-critical requirements, and scarce labeled data within a single, well-defined domain. Conversely, power systems can benefit from utilizing GML to complement classical solvers, as GML provide scalable, topology-aware approximations with promising generalization and computational efficiency. We identify open challenges, including limited real-world deployment and the need for interpretable models in safety-critical settings. Despite the rapidly growing number of publications, standardized benchmarks and open datasets remain scarce, leaving many results difficult to reproduce and undermining the long-term scientific credibility of the field. We further derive a structured requirements catalog for ML-ready power grid benchmarks, intended to guide future dataset development and improve reproducibility across studies. We call on the community to prioritize dedicated benchmark studies and the release of open datasets and models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-24c3be22ad6a7114", "title": "PertMind: Eliciting Emergent Biological Reasoning in LLM via Reinforcement Learning on Cellular Perturbation Data", "authors": ["Zhenchao Tang", "Xiaogang Xu", "Tianxu Lv", "Jiahui Guan", "Jiale Zhou", "Haohuai He", "Zhi Song", "Hanbo Huang", "Jiehui Huang", "Jiafei Wu", "Zhe Liu"], "year": 2026, "published": "2026-08-17T11:17:26Z", "updated": "2026-08-17T11:17:26Z", "arxiv_id": "http://arxiv.org/abs/2608.16419v1", "url": "https://arxiv.org/abs/2608.16419v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "q-bio.QM"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Chain-of-Thought", "Computer Use"], "benchmarks": [], "models": [], "key_results": ["Trained only on forward perturbation-response prediction, PertMind improved response inference in unseen cellular contexts while retaining general language capabilities."], "abstract": "Large language models can describe mechanisms, yet scalable post-training still depends on costly, manually curated biological reasoning traces. Here we show that cellular perturbation atlases can instead become reinforcement-learning environments, where measured gene responses provide computable rewards for biological reasoning. We introduce PertMind, which combines trusted-trajectory supervised initialization with gene-, pathway-, and format-level reinforcement signals. Trained only on forward perturbation-response prediction, PertMind improved response inference in unseen cellular contexts while retaining general language capabilities. It also transferred without task-specific post-training to reverse perturbation identification, double-perturbation reasoning, phenotypic-screen prioritization, and biological-process interpretation. PertMind further generated biological profiles that supported competitive gene, cell, and donor representations across multiscale downstream tasks. These results support the hypothesis that reinforcement on experimental endpoints can concentrate reusable biological strategies already accessible to pretrained models. More broadly, perturbation-derived reinforcement learning offers a scalable route for transforming expanding experimental atlases into training environments for general-purpose biological reasoning.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-c53a8d43e6114d08", "title": "POI Recommendation with LLM-Augmented Multi-Graph Learning and Contrastive Alignment", "authors": ["Burak Tamer", "Wolfram Höpken", "Zehui Wang"], "year": 2026, "published": "2026-08-17T11:01:12Z", "updated": "2026-08-17T11:01:12Z", "arxiv_id": "http://arxiv.org/abs/2608.16407v1", "url": "https://arxiv.org/abs/2608.16407v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": ["phic"], "key_results": ["Point-of-interest (POI) recommendation models based on graph neural networks achieve strong performance by propagating collaborative signals over user-item interactions, yet they struggle with the cold-start problem, where items with few or no interactions are not represented.", "Experiments on the Yelp Multimodal Recommendation Dataset show that LLM-MGCL outperforms classical collaborative filtering, matrix factorization, and interaction-only graph neural network baselines.", "It improves Recall@20 by 52.0% and NDCG@20 by 64.8% over LightGCN while performing on par with the strongest contrastive baseline, Self-supervised Graph Learning (SGL), which is also affected by the cold-start problem."], "abstract": "Point-of-interest (POI) recommendation models based on graph neural networks achieve strong performance by propagating collaborative signals over user-item interactions, yet they struggle with the cold-start problem, where items with few or no interactions are not represented. In this paper, we propose LLM-augmented Multi-Graph Contrastive Learning (LLM-MGCL), a multi-graph neural network that uses semantic and spatial information about items to extend the LightGCN backbone with two auxiliary item-item graphs: a semantic graph constructed from sentence embeddings of LLM-generated photo summaries and keywords, and a geographic graph derived from Haversine distances between business locations. Item embeddings are propagated over all three graphs in parallel, fused additively, and aligned across views through a bidirectional InfoNCE contrastive objective that connects behavioral, semantic, and spatial representations of the same items. Experiments on the Yelp Multimodal Recommendation Dataset show that LLM-MGCL outperforms classical collaborative filtering, matrix factorization, and interaction-only graph neural network baselines. It improves Recall@20 by 52.0% and NDCG@20 by 64.8% over LightGCN while performing on par with the strongest contrastive baseline, Self-supervised Graph Learning (SGL), which is also affected by the cold-start problem. An ablation study reveals that the cross-view contrastive alignment (CA) is the primary driver of these gains, with the best performance achieved when all three graphs are combined. Our results suggest that externally grounded, LLM-derived item knowledge can effectively compensate for missing collaborative signal and mitigate the item cold-start problem in POI recommendation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-f4806ec667d54aa9", "title": "HalluTracer: Hallucination Detection via Depth-Averaging Truth Signals", "authors": ["Zhihao Guo", "Zonghan Wu", "Huan Huo", "DaYong Ye", "Junwei Zhang", "Weiran Yao", "Zhiwei Liu", "Qingsong Wen", "Yilei Shao"], "year": 2026, "published": "2026-08-17T10:03:29Z", "updated": "2026-08-17T10:03:29Z", "arxiv_id": "http://arxiv.org/abs/2608.16353v1", "url": "https://arxiv.org/abs/2608.16353v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Across six open-source language models and five hallucination benchmarks, HalluTracer consistently outperforms matched white-box baselines, with gains ranging from one to fourteen points."], "abstract": "Even well-aligned large language models confidently generate factually incorrect text, making hallucination a persistent reliability risk in high-stakes deployments. These models nonetheless carry linearly separable truthfulness signals in their internal representations. Existing white-box detectors, however, collapse this evidence to isolated components or a single depth, discarding discriminative information distributed across the full forward pass. We introduce HalluTracer, a detection framework that reads and aggregates truthfulness evidence across every layer of the forward pass before the model emits any answer token. A geometric analysis reveals that the per-layer signals are weakly correlated, so that simple depth averaging suppresses layer-specific noise and captures nearly all linearly accessible information. Across six open-source language models and five hallucination benchmarks, HalluTracer consistently outperforms matched white-box baselines, with gains ranging from one to fourteen points. Collectively, our work recasts hallucination detection from a layer-selection problem into a depth-aggregation problem governed by the geometric sparsity of the truthfulness signal.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-8c5ff1107d2c6c36", "title": "Architecture-Dependent Causal Transfer of Activation States Across Large Language Models", "authors": ["Fernando Cardenas Piepereit"], "year": 2026, "published": "2026-08-17T09:53:27Z", "updated": "2026-08-17T09:53:27Z", "arxiv_id": "http://arxiv.org/abs/2608.16347v1", "url": "https://arxiv.org/abs/2608.16347v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["Qwen2-0.5B", "Mistral-7B", "Mistral-7B show no such effect despite comparable representational alignment at the hidden-state level. We interpret these results as evidence for causal transfer of the representational vehicle", "Phi-3-mini", "T5"], "key_results": ["Direct communication between AI systems relies on natural language as an intermediate layer, incurring encoding/decoding overhead, token cost, and latency."], "abstract": "Direct communication between AI systems relies on natural language as an intermediate layer, incurring encoding/decoding overhead, token cost, and latency. We ask whether internal activation states can instead be transferred causally between different large language model (LLM) architectures via a learned projection, evaluated at three levels: representational similarity, cross-model retrieval from projected states, and end-to-end causal transfer via activation injection during generation. Using four architecturally diverse open-weight models (Qwen2-0.5B, Phi-3-mini, Mistral-7B, FLAN-T5-base), we find that representational alignment in trained models exceeds a random-initialization null baseline and is best captured by a rank-based metric (mutual k-nearest-neighbour alignment), more robust to activation-magnitude outliers than centered kernel alignment (CKA) or Procrustes analysis. A learned projection network retrieves the correct target-model representation from a held-out set well above chance for the three causal decoder-only model pairs (45-50% top-1 accuracy vs. 5% chance) but at chance level for the encoder-based FLAN-T5. Injecting projected activations into a target model during generation produces a statistically significant, pre-registered causal effect on retrieval-based output similarity for only one of the three decoder-only pairs (Qwen2-0.5B to Phi-3-mini: 23.3% vs. 0.0% under negative control, p=0.047, FDR-corrected); the two pairs targeting Mistral-7B show no such effect despite comparable representational alignment at the hidden-state level. We interpret these results as evidence for causal transfer of the representational vehicle, not of meaning, and conclude that end-to-end activation-state transfer between LLMs, as currently implemented, is architecture-dependent rather than universal.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-821a5dc2e964a217", "title": "Task-Anchored Representation Shaping for Pre-Trained Model-Based Continual Learning", "authors": ["Zhiming Xu", "Huiyu Yi", "Zhen-Hao Xie", "Baile Xu", "Furao Shen", "Jian Zhao", "Suorong Yang"], "year": 2026, "published": "2026-08-17T09:51:21Z", "updated": "2026-08-17T09:51:21Z", "arxiv_id": "http://arxiv.org/abs/2608.16345v1", "url": "https://arxiv.org/abs/2608.16345v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Extensive experiments across multiple PTM-based continual learning paradigms show that TAILS can improve classification and task-inference performance with modest parameter overhead and negligible inference cost."], "abstract": "Pre-trained models (PTMs) provide a strong foundation for continual learning by offering stable representations that facilitate lightweight adaptation to new tasks. However, adapting well to each task does not ensure reliable inference over all learned tasks. Since task boundaries are often artificial and semantically entangled, an input from an unknown task can remain ambiguous even with strong PTM features, making cross-task prediction a key bottleneck. We propose Task-Anchored Inference Latent Shaping (TAILS), a lightweight post-PTM module that can be integrated into diverse continual learners and optimized through a decoupled step. TAILS uses fixed task anchors as persistent references to accumulated knowledge. It interprets each sample's feature representation relative to these references, then composes relevant evidence across tasks into latent recall. Rather than selecting a task-specific path or adjusting classifier outputs, TAILS uses latent recall to directly correct the feature representation before prediction. It therefore resolves cross-task ambiguity at the representation level, while leaving the original PTM, method-specific modules, and classifier unchanged. Extensive experiments across multiple PTM-based continual learning paradigms show that TAILS can improve classification and task-inference performance with modest parameter overhead and negligible inference cost.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-ee1dd7673f31379d", "title": "IndicQE-APE: A Benchmark for Quality Estimation and Automatic Post-Editing for Indic Languages", "authors": ["Diptesh Kanojia", "Archchana Sindhujan", "Sourabh Deoghare", "Daria Sokova", "Shenbin Qian", "Girish Koushik", "Tharindu Ranasinghe", "Constantin Orăsan", "Chrysoula Zerva", "Ricardo Rei", "Frédéric Blain", "André F. T. Martins", "Marco Turchi", "Matteo Negri", "Rajen Chatterjee", "Anoop Kunchukuttan", "Mitesh M. Khapra", "Pushpak Bhattacharyya"], "year": 2026, "published": "2026-08-17T09:50:52Z", "updated": "2026-08-17T09:50:52Z", "arxiv_id": "http://arxiv.org/abs/2608.16344v1", "url": "https://arxiv.org/abs/2608.16344v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Two of the axes are defined partly on the direct assessment and select a compressed slice of it, so each axis is compared against a control drawn from the same language pair with the same score distribution."], "abstract": "Indic quality estimation (QE) and automatic post-editing (APE) data is spread across separate releases, so no single resource supports training and evaluation across tasks and language pairs on one footing. We consolidate the WMT 2020--2024 shared-task lineage with an extended English--Malayalam resource into \\indicqe: $126{,}754$ instances over nine directional pairs, with up to four label types aligned on the same segment, a direct assessment, a human post-edit, word-level OK/BAD tags and an error explanation, and a test set stratified over four difficulty axes. On it, we benchmark six prompted LLMs and three COMET metrics on segment-level QE, and three systems on APE. Two of the axes are defined partly on the direct assessment and select a compressed slice of it, so each axis is compared against a control drawn from the same language pair with the same score distribution. Only one survives that control: segments whose holistic and token-level quality signals conflict are ranked worse than equally-scored segments of the same language, for all nine systems and all seven pairs that carry the axis. Annotator disagreement, which looks second-hardest without the control, has no effect with it. Few-shot prompting costs every model $\\leq$ $3.4$B both correlation and output-format compliance. Within-language accuracy does not make scores comparable across pairs: of the three trained metrics, the one with the best within-language correlation loses most when the pairs are pooled. The benchmark and code will be released.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-6e95153a234dff0a", "title": "LiD-GLM: Lipschitz-constrained Deep Generalized Linear Models", "authors": ["Tom Splittgerber", "Niklas Koenen", "Marvin N. Wright", "Werner Brannath"], "year": 2026, "published": "2026-08-17T09:45:35Z", "updated": "2026-08-17T09:45:35Z", "arxiv_id": "http://arxiv.org/abs/2608.16340v1", "url": "https://arxiv.org/abs/2608.16340v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["GLM"], "key_results": ["The combination of traditional statistical models and neural network (NN) components into semi-structured hybrid models is an intriguing approach to construct models that, ideally, combine traditional interpretability with the unprecedented flexibility of NNs."], "abstract": "The combination of traditional statistical models and neural network (NN) components into semi-structured hybrid models is an intriguing approach to construct models that, ideally, combine traditional interpretability with the unprecedented flexibility of NNs. In order to preserve interpretability, it is usually necessary to restrict the NN components to prevent them from dominating the model. However, existing methods that enforce structural constraints on their NN components severely limit their models' flexibility; in contrast, methods that only enforce weak, indirect constraints lose meaningful interpretability. The method we propose therefore leverages invertible residual neural networks (i-ResNets) to equip generalized linear models with both nonlinear parameter estimation and a flexible correction of their distributional assumptions while always retaining stochastic monotonicity of the modeled distribution in the (formerly linear) predictor. The i-ResNets correspond to a controlled deviation from identity and by constraining their Lipschitz constant one can rigorously limit and quantify how far the hybrid model deviates from its traditional counterpart. This enables a user-specifiable compromise between flexibility and interpretability without limiting the structure of nonlinear and interaction effects that can be learned. Furthermore, we develop specific inherent interpretation techniques for our model and enforce model identifiability through an adapted post-hoc orthogonalization.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-e2abeee30c4bada8", "title": "Toward Decentralized Carbon Trading in Indonesia: A Public-Blockchain Architecture for Tokenized Real-World Assets", "authors": ["Rischan Mafrur", "Fadli Ikhsan Pratama", "Khadijah"], "year": 2026, "published": "2026-08-16T07:34:39Z", "updated": "2026-08-16T07:34:39Z", "arxiv_id": "http://arxiv.org/abs/2608.15597v1", "url": "https://arxiv.org/abs/2608.15597v1", "primary_category": "cs.CE", "categories": ["cs.CE", "q-fin.CP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Indonesia has established a regulated carbon market supported by national registry infrastructure and the IDXCarbon exchange."], "abstract": "Indonesia has established a regulated carbon market supported by national registry infrastructure and the IDXCarbon exchange. Carbon units can be issued, recorded, traded, and retired within this framework. IDXCarbon currently uses a private blockchain for its trading infrastructure. This creates an opportunity to examine how Indonesian carbon credits could also be represented and traded through public blockchain infrastructure.\n  This study proposes an architecture for tokenizing Indonesian carbon credits as real-world assets (RWAs), with particular focus on Sertifikat Pengurangan Emisi Gas Rumah Kaca (SPE-GRK). The proposed architecture retains the Sistem Registri Unit Karbon (SRUK) as the authoritative source of carbon-unit status. It introduces a public-blockchain layer for token representation and programmable transactions. The architecture is designed to support lifecycle management, token-based asset representation, public observability of token activity, interoperability, wallet-based transactions, and programmable settlement.\n  The architecture consists of four layers: the authoritative carbon layer, the registry interoperability and tokenization layer, the public-blockchain RWA layer, and the market and application layer. Access to the tokenized carbon assets remains regulated. Token issuance and transfers are linked to participant eligibility and registry status. Retirement also remains dependent on the authoritative carbon registry. The proposed architecture provides a framework for introducing public-blockchain RWA infrastructure into Indonesia's existing carbon market while maintaining SRUK authority and existing market-integrity controls.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-e512188af87b82e1", "title": "Amortised Post-Hoc Explanation with Exact Preservation for Dynamic Graph Anomaly Detectors", "authors": ["Iyad Assaad Nekka", "Hamida Seba", "Walid Khaled Hidouci", "Karima Amrouche"], "year": 2026, "published": "2026-08-16T06:10:40Z", "updated": "2026-08-16T06:10:40Z", "arxiv_id": "http://arxiv.org/abs/2608.15559v1", "url": "https://arxiv.org/abs/2608.15559v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Interpretability", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["X-StrGNN attains the highest stability (0.913) at 268x lower cost than per-instance optimisation, and its temporal attribution (1.601 against a measured random floor of 0.973) is separably better than its ablated control, while per-instance optimisation - the most expensive strategy - falls below that floor."], "abstract": "Anomaly detection in dynamic graphs underpins financial fraud analysis, intrusion detection, and platform integrity, where automated decisions require human-interpretable justifications. StrGNN, the strongest performer in recent benchmarks, produces no explanation: when an edge is flagged, the analyst receives only a score. Explanation metrics are undefined for StrGNN because no attribution vector exists. This paper closes that gap. We present X-StrGNN, a post-hoc explanation layer that wraps a trained, frozen StrGNN and emits, for every flagged edge, dual attributions: a structural attribution identifying which contextual interactions in the enclosing subgraph drove the decision, and a temporal attribution identifying which historical snapshot carried the signal. Both attributions are multiplicative masks identically one in the unexplained pass, so the layer is an exact pass-through: detection is preserved to machine precision, verified rather than asserted (Delta AUC = 0.0000, Delta AP = 0.0000, Delta P@100 = 0.0000). Attribution costs 0.66 ms per edge, making explanation of an entire alarm list feasible. We conduct the first controlled design study of attribution strategies for this architecture, comparing gradient attribution, per-instance mask optimisation, and amortised parameterisation under one protocol, one budget, and three seeds. X-StrGNN attains the highest stability (0.913) at 268x lower cost than per-instance optimisation, and its temporal attribution (1.601 against a measured random floor of 0.973) is separably better than its ablated control, while per-instance optimisation - the most expensive strategy - falls below that floor. Code, protocol, and per-seed measurements are released.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-df1e1cfa4ad19d65", "title": "Spectral Saliency for Machine Unlearning", "authors": ["Cedar Site Bai", "Amber Yijia Zheng", "Raymond A. Yeh", "Brian Bullins"], "year": 2026, "published": "2026-08-16T05:44:53Z", "updated": "2026-08-16T05:44:53Z", "arxiv_id": "http://arxiv.org/abs/2608.15548v1", "url": "https://arxiv.org/abs/2608.15548v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Machine unlearning (MU) aims to remove the influence of specific training data while preserving model utility."], "abstract": "Machine unlearning (MU) aims to remove the influence of specific training data while preserving model utility. As the name suggests, MU can be viewed as the inverse of learning, using gradient-based updates to reduce the influence of a forget-set by counteracting the previously learned behavior. Recently, Muon, a gradient descent variant, has been introduced. Muon applies spectral magnitude normalization to encourage exploration of rare directions and demonstrates promising performance. Inspired by Muon, we adopt the spectral view for unlearning and propose Spectral Saliency Unlearning (SSU). SSU thresholds weak singular components and updates only those directions supported by a confident unlearning signal. We further provide theoretical justification for this thresholding approach from the perspective of the forgetting-retention trade-off. Experiments across image classifiers, diffusion models, and LLMs demonstrate SSU's effectiveness.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-c022bb505f295fce", "title": "Not All Attention Is Equal: A Quantitative Survey of the EEI Trade-off", "authors": ["Aditya Singh"], "year": 2026, "published": "2026-08-16T00:37:38Z", "updated": "2026-08-16T00:37:38Z", "arxiv_id": "http://arxiv.org/abs/2608.15459v1", "url": "https://arxiv.org/abs/2608.15459v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Preference Optimization", "Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["Attention mechanisms have driven machine learning for a decade, from neural machine translation to language models that do general-purpose reasoning."], "abstract": "Attention mechanisms have driven machine learning for a decade, from neural machine translation to language models that do general-purpose reasoning. This survey covers four connected threads: their formulation for sequence-to-sequence tasks, adaptation to computer vision, efficiency innovations that address the quadratic bottleneck, and advances in interpretability. We define three criteria: efficiency, expressiveness, and interpretability, and compare twenty-one methods using an EEI scoring framework. Scores come from a single rater with an assumed +/-1-point perturbation range. A deterministic Monte Carlo analysis with 200,000 samples shows that, under this perturbation model, rank changes of more than one position occur in 67-70% of samples on average. A rank-matched null model reproduces a similar stability profile, so the results support coarse tier-level comparisons rather than fine-grained rankings. The survey traces attention from Bahdanau-Luong alignment through the Transformer and into vision architectures. It reviews fixed and learned sparse attention, linear attention, IO-aware exact algorithms including FlashAttention, and state-space alternatives including Mamba. It also covers induction heads, superposition, and the attention-SSM duality. We further provide a structured narrative review, a benchmark synthesis with cross-study caveats, a five-problem research gap analysis, and a 2015-2026 evolution timeline. We conclude by framing attention research as an expansion of the efficiency-expressiveness-interpretability frontier and identifying future directions including unified efficiency benchmarks, learned routing for hybrid architectures, length generalization, and scalable mechanistic interpretability.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-3f6128928662b6ee", "title": "Detecting Money Laundering in Rwandan Mobile Money: A Machine Learning Framework", "authors": ["Emmanuel Nahimana", "Yaé Ulrich Gaba"], "year": 2026, "published": "2026-08-15T23:19:34Z", "updated": "2026-08-15T23:19:34Z", "arxiv_id": "http://arxiv.org/abs/2608.15447v1", "url": "https://arxiv.org/abs/2608.15447v1", "primary_category": "cs.LG", "categories": ["cs.LG", "q-fin.RM"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Mobile money has widened financial access across Sub-Saharan Africa and enlarged the surface for money-laundering and terrorism-financing (ML/TF) activity in ecosystems dominated by high-volume, low-value transactions."], "abstract": "Mobile money has widened financial access across Sub-Saharan Africa and enlarged the surface for money-laundering and terrorism-financing (ML/TF) activity in ecosystems dominated by high-volume, low-value transactions. Rwanda is a case in point: several million active mobile-money users, telecom-led wallets on the MTN and Airtel networks, and a Financial Intelligence Centre (FIC) supervising transaction streams whose scale exceeds static rule-based monitoring. This paper develops and evaluates a transaction-monitoring framework aligned to the Rwandan AML/CFT regime under (i) extreme class imbalance (~0.1% prevalence), (ii) scarce and delayed labels, and (iii) bounded investigator capacity. Using SAML-D, a synthetic dataset of 9,504,852 transactions with 17 laundering typologies, we engineer account-centric behavioural features (rolling velocity, net-flow directionality, counterparty diversity, burstiness) and benchmark supervised classifiers (Logistic Regression, Random Forest, LightGBM), unsupervised anomaly detectors (Isolation Forest, Local Outlier Factor), a dense autoencoder, and a late-fusion meta-learner. Evaluation is operational: PR-AUC, recall at a calibrated ~90%-precision point, recall at top-K%, and alerts per 10,000. On the chronologically held-out test period, LightGBM attains PR-AUC = 0.0469, capturing 64 laundering cases at precision ~0.89 with 0.51 alerts per 10,000; the fusion stacker reaches PR-AUC = 0.0477 at precision ~0.91 and 0.46 alerts per 10,000, recovering 59 true positives. We map score bands to Rwanda-relevant analyst workflows and STR/SAR escalation, and outline a staged path from synthetic prototyping to real-data validation with the National Bank of Rwanda and FIC. The contribution is operational: a governance-aware pipeline and evaluation protocol calibrated to the constraints of an African mobile-money regulator, not a new algorithm.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-0ddc415cde701e4f", "title": "NeuRoute: Logit-Guided Neural Routing for Billion-Scale Vector Search with Sub-Hour Index Construction", "authors": ["Xingqiao Wang", "Zi Wang", "Xiaowei Xu"], "year": 2026, "published": "2026-08-15T22:44:10Z", "updated": "2026-08-15T22:44:10Z", "arxiv_id": "http://arxiv.org/abs/2608.15438v1", "url": "https://arxiv.org/abs/2608.15438v1", "primary_category": "cs.DB", "categories": ["cs.DB", "cs.IR", "cs.LG"], "pillar": "Quant × AI", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["On billion-scale benchmarks, NeuRoute achieves strong accuracy-throughput trade-offs with fast index construction: on BigANN-1B it reaches $90.3\\%$ Recall@10 at 2,414 QPS and is $1.7\\times$ faster than OPQ+IVF-PQ (refine) at comparable accuracy, while completing end-to-end training+construction in under an hour on both BigANN-1B and Deep1B-1B."], "abstract": "Building approximate nearest neighbor (ANN) indexes at billion scale is often dominated by expensive global clustering or graph construction, making time-to-index a first-order systems concern. We present NeuRoute, a learned hashing index that turns short binary codes into an effective routing primitive for large-scale vector search. NeuRoute trains a lightweight neural network encoder with a selective similarity-preserving objective to produce well-balanced binary addresses. During construction, NeuRoute organizes vectors into buckets by their codes and performs bucket-local clustering in the encoder's low-dimensional space to form centroids. At query time, NeuRoute exploits the encoder logits as an uncertainty signal: it uses deviation-to-threshold scores to prioritize uncertain-bit perturbations for query-adaptive multi-bucket probing, scores bucket-local centroids by their distances to the query to form a compact candidate cluster set, and applies centroid-stage gating with heap-quality-driven early stopping to prune low-value clusters before exact refinement. On billion-scale benchmarks, NeuRoute achieves strong accuracy-throughput trade-offs with fast index construction: on BigANN-1B it reaches $90.3\\%$ Recall@10 at 2,414 QPS and is $1.7\\times$ faster than OPQ+IVF-PQ (refine) at comparable accuracy, while completing end-to-end training+construction in under an hour on both BigANN-1B and Deep1B-1B. These results show that logit-guided neural routing can make hashing competitive as a lightweight ANN indexing framework at billion scale. Source code and artifacts are available at https://github.com/XingqiaoWang/NeuRoute.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-895f3f9648dc4676", "title": "SAGA: Structure-Attended Generative Action Embedding Model that encodes Multi-Surface User Action Sequences", "authors": ["Tsz Fung Pang", "Po Jen Chen", "Nimish Ronghe", "Farhad Farahani", "Bo Zhang"], "year": 2026, "published": "2026-08-15T22:11:34Z", "updated": "2026-08-15T22:11:34Z", "arxiv_id": "http://arxiv.org/abs/2608.15429v1", "url": "https://arxiv.org/abs/2608.15429v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.IR"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Prior embedding models for sequential recommendation typically operate within a homogeneous action space, limiting their ability to capture cross-surface behavioral signals spanning distinct behavioral domains."], "abstract": "Prior embedding models for sequential recommendation typically operate within a homogeneous action space, limiting their ability to capture cross-surface behavioral signals spanning distinct behavioral domains. We present SAGA, a generative action embedding model that encodes multi-surface user interaction sequences across a Financial Service organization's ecosystems, from checkout, peer-to-peer (P2P) transactions, in-app engagement, email to account actions, into a unified user representation for downstream recommendation tasks. Central to SAGA is a per-field tokenization schema that decomposes each action event into multiple field-level tokens (e.g. product, interaction, surface), enabling field-level attention and per-field training objectives that fused single-token approaches cannot support. Through an offline ablation study on loss formulation, tokenization granularity and training data scope, we isolate the contribution of each design choice. A downstream model integrated with SAGA-generated user embeddings delivers the strongest overall click and conversion lift across diverse downstream touchpoints, compared to all ablated and alternative architectures.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-6d77ceaca79ef5ce", "title": "SAPE: Sandwich Adapters for Parameter Efficiency in Large Language Model Fine-Tuning", "authors": ["Mohammad Aref Jafari-Raddani", "Morteza Mohajjel Kafshdooz"], "year": 2026, "published": "2026-08-15T18:32:47Z", "updated": "2026-08-15T18:32:47Z", "arxiv_id": "http://arxiv.org/abs/2608.15360v1", "url": "https://arxiv.org/abs/2608.15360v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Memory / RAG"], "benchmarks": ["GSM8K"], "models": ["LLaMA-3.2", "yielding", "BERT"], "key_results": ["While Parameter-Efficient Fine-Tuning (PEFT) has substantially reduced the hardware cost of adapting Large Language Models (LLMs) by decreasing the number of trainable parameters, recent studies have sought to further improve PEFT through parameter sharing.", "This design significantly reduces memory consumption while eliminating the computational overhead associated with dynamic parameter-sharing methods.", "Extensive evaluations across encoder-only and causal decoder architectures demonstrate that SAPE achieves state-of-the-art performance in low-parameter regimes."], "abstract": "While Parameter-Efficient Fine-Tuning (PEFT) has substantially reduced the hardware cost of adapting Large Language Models (LLMs) by decreasing the number of trainable parameters, recent studies have sought to further improve PEFT through parameter sharing. However, these approaches either employ uniform parameter sharing across layers, which can delay convergence, or rely on dynamic masking strategies, which add computational overhead. The potential of sharing patterns inspired by the inherent hierarchical structure of Transformer architectures remains unexplored in PEFT. To address this gap, we introduce SAPE (Sandwich Adapters for Parameter Efficiency), a PEFT framework based on a sandwich-style hard weight-sharing topology. SAPE routes intermediate Transformer layers through balanced shared group adapters while strictly isolating the input embedding and final projection boundary transformations to prevent gradient interference. This design significantly reduces memory consumption while eliminating the computational overhead associated with dynamic parameter-sharing methods. Extensive evaluations across encoder-only and causal decoder architectures demonstrate that SAPE achieves state-of-the-art performance in low-parameter regimes. On natural language understanding, SAPE outperforms proPETL on RoBERTa-large while utilizing only 10% of the baseline's parameter budget. On natural language generation and world knowledge reasoning with LLaMA-3.2 (3B) under a strict ~0.6M parameter constraint, SAPE outperforms AdaLoRA, yielding absolute improvements of +4.85% on GSM8K and +3.11% on CommonsenseQA. Furthermore, through comprehensive topological ablations, we formalize an inherent capacity trade-off: while hard parameter sharing strongly regularizes semantic generalization, it slightly smooths the sharp layer-wise transformations required for rigid multi-step arithmetic reasoning.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-58d180a2bc54667c", "title": "Decentralized Federated Learning for Heterogeneous Multi-Task Semantic Communication", "authors": ["Lin Yin", "Tiejun Lv", "Weicai Li", "Xi Yu", "Xiaoyu He"], "year": 2026, "published": "2026-08-15T14:22:58Z", "updated": "2026-08-15T14:22:58Z", "arxiv_id": "http://arxiv.org/abs/2608.15256v1", "url": "https://arxiv.org/abs/2608.15256v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": ["yields"], "key_results": ["We reveal a strict Ushaped trade-off: deeper topological mixing reduces variance but amplifies structural OCB.", "At the analytically derived optimal aggregation depth, our method achieves a 4.77% global relative improvement over the no-aggregation baseline and outperforms decentralized FedAvg, FedAMP, and heuristic max aggregation."], "abstract": "Collaborative training in distributed semantic communication (DSC) networks typically relies on decentralized federated learning (DFL). However, pushing topology-agnostic aggregation into heterogeneous, multi-task environments creates a fundamental bottleneck: it drives negative transfer and overconsensus bias (OCB). This paper introduces a personalized DSC framework that cuts off this cross-task interference. At the node level, a policy-driven multi-path routing mechanism separates task-specific features from shared representations to preserve local fidelity. Across the network, we deploy a \"communicationwhile- aggregation\" protocol. It calibrates a column-stochastic consensus matrix using task affinities. This limits the system to absorbing complementary knowledge while actively blocking mismatched parameter updates. To bound the convergence, we derive a unified Lyapunov drift analysis. We reveal a strict Ushaped trade-off: deeper topological mixing reduces variance but amplifies structural OCB. Resolving this tension yields a closed-form expression for the optimal aggregation depth. We evaluate the proposed framework on NYU-v2, where the results reveal a clear trade-off between insufficient aggregation and excessive topological mixing. At the analytically derived optimal aggregation depth, our method achieves a 4.77% global relative improvement over the no-aggregation baseline and outperforms decentralized FedAvg, FedAMP, and heuristic max aggregation. We further evaluate the framework on Taskonomy and imperfect wireless links to examine the effects of network-size variation and wireless-link reliability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-38a092a0254c3f0d", "title": "FinFraudBench: A Heterogeneous Graph Benchmark for Financial Fraud Detection", "authors": ["Yixuan Chen", "Hongyu Zhan", "Jie Sheng", "Weiyu Han", "Shuai Chen", "Tianyi Zhang", "Xiao Tan", "Jun Xia"], "year": 2026, "published": "2026-08-15T11:34:51Z", "updated": "2026-08-15T11:34:51Z", "arxiv_id": "http://arxiv.org/abs/2608.15177v1", "url": "https://arxiv.org/abs/2608.15177v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["yield"], "key_results": ["The increasing complexity of digital financial systems has reshaped financial fraud detection from isolated transaction classification into relational risk reasoning over interconnected financial entities."], "abstract": "The increasing complexity of digital financial systems has reshaped financial fraud detection from isolated transaction classification into relational risk reasoning over interconnected financial entities. This shift has motivated graph-based fraud detection, where models identify fraudulent nodes by exploiting dependencies among customers, cards, merchants, categories, and locations. However, despite rapid progress in graph-based methods, existing public benchmarks remain misaligned with real-world financial systems in two important aspects. First, they often simplify financial ecosystems into homogeneous or single-node-type multi-relational graphs, failing to preserve the multi-entity and multi-relational nature of financial data. Second, they rarely provide large-scale heterogeneous financial graph datasets with realistic operating conditions such as extreme class imbalance and limited label availability, making it difficult to assess the practical effectiveness of current methods. To address these gaps, we present FinFraudBench, a heterogeneous graph benchmark for financial fraud detection. FinFraudBench contains two heterogeneous graph datasets (CreditCard-Fraud and BankTrans-Fraud) with up to 8.99M nodes and 89.23M directed typed edges. Each dataset preserves six financial entity types, fourteen directed edge types, and natural fraud rates that mirror deployment constraints. With these datasets, we establish a standardized evaluation protocol covering both ranking and imbalance-sensitive classification metrics, and evaluate representative baselines. Extensive experiments yield empirical insights into current methods' limitations and suggest promising avenues for future research. FinFraudBench is available at https://anonymous.4open.science/r/FinFraudBench-B002.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-6cebec8ce5323fa2", "title": "An Adaptive Gradient Clipping and Noise Injection Mechanism for Differentially Private Federated Learning", "authors": ["Wenjing Wei", "Alla Jammine", "Farid Nait-Abdesselam"], "year": 2026, "published": "2026-08-15T10:10:34Z", "updated": "2026-08-15T10:10:34Z", "arxiv_id": "http://arxiv.org/abs/2608.15153v1", "url": "https://arxiv.org/abs/2608.15153v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Differentially private federated learning must balance privacy protection against model accuracy and training efficiency.", "Compared with the static DDP-SA baseline, DDP-SA-adaptive reduces the number of communication rounds by 6.81%, total training time by 19.21%, and average per-round training time by 13.33%, leading to improved training efficiency.", "It also reduces test loss by 98.74% and increases test R2 by 3.41%, leading to improved model accuracy."], "abstract": "Differentially private federated learning must balance privacy protection against model accuracy and training efficiency. Static gradient clipping applies a fixed threshold throughout training and across model layers, which can cause excessive clipping when the threshold is too small or unnecessarily large noise when it is too large. This paper presents DDP-SA-adaptive, an adaptive gradient clipping and noise adding mechanism for differentially private federated learning with secure aggregation. At each communication round, every client determines a separate clipping threshold for each model layer from the median of its per-sample gradient norms. The resulting layer-wise thresholds adapt to the evolving gradient distributions and calibrate the Laplace noise added before the updates are encoded and secret-shared among intermediate aggregation servers. We evaluate the proposed mechanism on a federated regression task in terms of efficiency, accuracy, privacy, convergence, clipping norm, and noise magnitude. Compared with the static DDP-SA baseline, DDP-SA-adaptive reduces the number of communication rounds by 6.81%, total training time by 19.21%, and average per-round training time by 13.33%, leading to improved training efficiency. It also reduces test loss by 98.74% and increases test R2 by 3.41%, leading to improved model accuracy. To attain R2 = 0.99, the adaptive mechanism operates with a privacy budget of approximately epsilon = 0.1, compared with epsilon = 0.4 for static DDP-SA, thus providing stronger privacy protection and achieving stronger privacy guarantees. These results demonstrate that round-wise, layer-wise adaptation can improve the privacy-accuracy-efficiency trade-off of differentially private federated learning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-78a1111b9ad0614c", "title": "Decision-Driven Regularization: A Blended Model for Learning and Optimization", "authors": ["Gar Goei Loke", "Qinshen Tang", "Yangge Xiao", "Xun Zhang"], "year": 2026, "published": "2026-08-15T08:49:33Z", "updated": "2026-08-15T08:49:33Z", "arxiv_id": "http://arxiv.org/abs/2608.15124v1", "url": "https://arxiv.org/abs/2608.15124v1", "primary_category": "cs.LG", "categories": ["cs.LG", "math.OC"], "pillar": "Quant × AI", "methods": ["Deep Research", "Quant / Trading"], "benchmarks": ["MATH"], "models": [], "key_results": ["We focus on the integrated learning and optimization literature, and identify that a lack of control for prediction accuracy can lead to overfitting and a loss of decision effectiveness against simple separate learning and optimization models.", "Our model is shown to be numerically superior to other benchmarks, such as OLS, Random Forest, XGBoost, SPO+, Perturbation Gradient, and Learning and Rank, in our synthetic studies."], "abstract": "In contextual optimization, the decision-maker seeks optimal decisions to minimize a cost function, that varies based on observed features. This context is common in many business applications ranging from on-demand delivery and retail operations to portfolio optimization and inventory management. In this paper, we study the learning and optimization approach, which first learns how outcomes result from the features, and then selects optimal decisions based on these outcomes. We focus on the integrated learning and optimization literature, and identify that a lack of control for prediction accuracy can lead to overfitting and a loss of decision effectiveness against simple separate learning and optimization models. Instead, we propose a bi-objective formulation that balances prediction accuracy and cost minimization, termed decision-driven regularization. It also addresses ambiguity in the definition of the cost function via a surrogate that depends on a new hyperparameter. We additionally show that alternative perspectives for formulating the problem, namely robust optimization and regret minimization, lead to models that are closely related to our proposed model. As a consequence, our framework generalizes models such as SPO+. Our model is shown to be numerically superior to other benchmarks, such as OLS, Random Forest, XGBoost, SPO+, Perturbation Gradient, and Learning and Rank, in our synthetic studies.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-0993b2f20aa26c3e", "title": "RecurrentGPT: Expressive Depth through Recurrent Modulation in Transformers", "authors": ["Amr Hegazy", "Amr Alanwar", "Mostafa Elhoushi"], "year": 2026, "published": "2026-08-15T06:22:00Z", "updated": "2026-08-15T06:22:00Z", "arxiv_id": "http://arxiv.org/abs/2608.15062v1", "url": "https://arxiv.org/abs/2608.15062v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "Quant × AI", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["This allows the model to specialize the input to the same few layers across recurrences, rather than requiring many unique layers to achieve functional diversity.", "Under an isoFLOPS constraint, a 3-layer RecurrentGPT matches the accuracy of a 12-layer GPT-2 Small baseline with similar training and inference FLOPs, and leads MoR and heavy-tail depth sampling in all nine scale-by-budget cells; at medium and large scale it approaches dense quality at the standard token budget and overtakes it at medium scale once that budget is doubled.", "Under an isoPARAMS constraint, deeper recurrence achieves a 2.76 validation loss versus 2.84 for a non-recurrent counterpart at matched parameter and data budget."], "abstract": "Scaling transformer language models creates an inherent tension between expressivity and memory efficiency. While unique weights across layers preserve functional specialization---from input-grounding to abstract refinement---they incur a substantial memory footprint. Conversely, standard depth-sharing enforces uniform transformations that collapse representational diversity and degrade modeling quality. We introduce RecurrentGPT, a recurrent depth transformer where fixed-depth prelude and coda blocks bracket a single shared core iterated R times. Inspired by gated recurrent neural networks, we employ a lightweight projection and an elementwise update gate---conditioned on the hidden state, the fixed prelude output, and noise resampled at every step---to modulate the recurrent update. This allows the model to specialize the input to the same few layers across recurrences, rather than requiring many unique layers to achieve functional diversity. Under an isoFLOPS constraint, a 3-layer RecurrentGPT matches the accuracy of a 12-layer GPT-2 Small baseline with similar training and inference FLOPs, and leads MoR and heavy-tail depth sampling in all nine scale-by-budget cells; at medium and large scale it approaches dense quality at the standard token budget and overtakes it at medium scale once that budget is doubled. Under an isoPARAMS constraint, deeper recurrence achieves a 2.76 validation loss versus 2.84 for a non-recurrent counterpart at matched parameter and data budget. Our results demonstrate that adaptive depth reuse is a principled strategy for trading parameters for quality: at large scale, 63% fewer parameters and 59% less peak decoding memory for a 10% increase in compiled generation latency.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-16836495397fa7c7", "title": "Orbit-Planner: Towards Latent World Models for On-Orbit Obstacle Avoidance of Satellite Agents", "authors": ["Zhijian Li", "Chao Ren", "Peijin Wang", "Xian Sun"], "year": 2026, "published": "2026-08-17T14:49:07Z", "updated": "2026-08-17T14:49:07Z", "arxiv_id": "http://arxiv.org/abs/2608.16651v1", "url": "https://arxiv.org/abs/2608.16651v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["World Model"], "benchmarks": [], "models": [], "key_results": ["Satellite agents for on-orbit navigation tasks need to predict collision risks using limited onboard observations."], "abstract": "Satellite agents for on-orbit navigation tasks need to predict collision risks using limited onboard observations. However, conventional planners often rely on predefined maps and fixed environmental assumptions, limiting their adaptability in dynamic on-orbit scenarios. In this paper, we propose Orbit-Planner, a two-stage latent world model for on-orbit obstacle avoidance. Orbit-Planner learns action-conditioned spacecraft dynamics to perform future-state rollouts in latent space, and introduces a Physics Probe to decode physical state changes from imagined latent trajectories. Experiments demonstrate that Orbit-Planner can perform long-horizon latent rollouts and recover physical states from imagined trajectories. In closed-loop obstacle-avoidance navigation in Isaac Sim, it attains a success rate of 91.7%. Code is available at https://github.com/ZhijianLi2003/Orbit_Planner.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 18080, "fulltext_methods": ["Test-time Scaling", "Long Context", "Interpretability", "Safety / Jailbreak", "World Model", "Video Generation", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["ying"], "matched_methods": ["World Model"], "checked_on": "2026-08-19T03:31:20.113668+00:00"}}
{"id": "evg-0aab611a0398e07c", "title": "PDDLCoder: Agentic PDDL Generation for LLM-Assisted Symbolic Planning", "authors": ["Veit Laule", "Jiangtao Shuai", "Manfred Hauswirth", "Sonja Schimmler"], "year": 2026, "published": "2026-08-17T14:34:56Z", "updated": "2026-08-17T14:34:56Z", "arxiv_id": "http://arxiv.org/abs/2608.16637v1", "url": "https://arxiv.org/abs/2608.16637v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["This improves upon our adaptations of previous PDDL generation methods, which achieved up to 45.3\\%, and outperforms direct LLM planning approaches, which reached up to 74.5\\% on the same test set."], "abstract": "LLMs remain unreliable for long-horizon planning, often generating logically inconsistent or non-applicable plans. Recent hybrid methods instead translate natural language into the Planning Domain Definition Language (PDDL), allowing symbolic planners to produce verifiable plans. However, existing methods frequently rely on rigid generation pipelines, a partial PDDL definition, or human feedback. Furthermore, their evaluation is hindered by the lack of standardized benchmarks with automated verification. To address these limitations, we present PDDLCoder, an agentic framework for PDDL generation from natural language that iteratively generates, analyzes, and refines planning specifications. We further introduce NL-pddlgym, a benchmark dataset comprising 711 planning problems across 23 domains with executable gym environments for the automated verification of plan applicability. Experiments on the NL-pddlgym test set containing 106 problems across 4 held-out domains show that PDDLCoder generates applicable plans for 89.6\\% of tested planning problems. This improves upon our adaptations of previous PDDL generation methods, which achieved up to 45.3\\%, and outperforms direct LLM planning approaches, which reached up to 74.5\\% on the same test set. Our work demonstrates the effectiveness of agentic PDDL generation for planning and establishes a reproducible benchmark for future research on LLM-assisted symbolic planning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00", "verified": true, "verification": {"source": "arxiv-html", "status": "fulltext-verified", "fulltext_chars": 76594, "fulltext_methods": ["RLVR / GRPO", "Chain-of-Thought", "Test-time Scaling", "Long Context", "Safety / Jailbreak", "Multi-Agent", "Memory / RAG", "World Model", "VLM"], "fulltext_benchmarks": [], "fulltext_models": ["Llama 4", "llama-4-multimodal-intelligence", "Qwen", "Qwen3.6-35B-A3B", "qwen.ai", "qwen3.6-35b-a3b", "DeepSeek", "Deepseek", "Deepseek-v4", "r1"], "matched_methods": ["Multi-Agent"], "checked_on": "2026-08-19T03:31:22.752711+00:00"}}
{"id": "evg-98ce694faf17975c", "title": "Cost Scales with Change, Not Corpus Size: Incrementally Maintaining an Evolving Semantic Substrate", "authors": ["Yusuke Takahashi", "Kyle Wild", "Asako Uraki"], "year": 2026, "published": "2026-08-17T14:21:15Z", "updated": "2026-08-17T14:21:15Z", "arxiv_id": "http://arxiv.org/abs/2608.16621v1", "url": "https://arxiv.org/abs/2608.16621v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.DB", "cs.IR"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Retrieval-augmented and agentic question-answering systems increasingly re-derive the meaning of a corpus at query time."], "abstract": "Retrieval-augmented and agentic question-answering systems increasingly re-derive the meaning of a corpus at query time. Put plainly, instead of re-deriving what a corpus means on every question, the work is done once when a document arrives and is thereafter merely consulted -- a compiler, not an interpreter, of meaning. An alternative is to compile that meaning once, at ingest time, into a compact, queryable semantic substrate and maintain it as the corpus evolves. The central objection is maintenance cost: rebuilding a truncated singular value decomposition (SVD) on every change appears prohibitive, and a change of embedding model seems to force a full re-embedding. We argue and show empirically that maintenance cost scales with the amount of change, not corpus size. On a controlled synthetic pilot (dimension 256, rank 32, a corpus grown from 3,000 to 9,000 documents over 50 update events), incremental low-rank updates were 33.7 times cheaper per update than full re-SVD and 23.8 times cheaper cumulatively, while the incremental subspace tracked the full recomputation to within floating-point precision (maximum principal-angle drift below 1e-11 degrees; recall@10 = 1.0). An orthogonal Procrustes virtual axis update recovered 0.95 mean cosine to truly re-embedded vectors by re-embedding only about 10 percent of the corpus. The results support maintaining, rather than repeatedly reconstructing, a semantic substrate.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-e7ea1b3a3d5c5af7", "title": "\"If It Looks Like a User\": Measuring Real-Time Moderation Effects via Social Media Simulation", "authors": ["Enrico Verdolotti", "Gianluca Nogara", "Luca Luceri", "Silvia Giordano"], "year": 2026, "published": "2026-08-17T13:58:51Z", "updated": "2026-08-17T13:58:51Z", "arxiv_id": "http://arxiv.org/abs/2608.16601v1", "url": "https://arxiv.org/abs/2608.16601v1", "primary_category": "cs.SI", "categories": ["cs.SI", "cs.CY", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Our approach replaces ad-hoc parametrisations with empirically fitted distributions, optimised via CMA-ES (Covariance Matrix Adaptation Evolution Strategy) and validated against real data across temporal, distributional, and structural dimensions."], "abstract": "Agent-based social media simulators offer a controlled environment to study content moderation, yet their value hinges on how faithfully they reproduce real platform dynamics. We develop a calibrated extension of SimSoM, an agent-based model of information diffusion on social networks, grounded in a real-world dataset of online vaccine discourse during the COVID-19 pandemic. Our approach replaces ad-hoc parametrisations with empirically fitted distributions, optimised via CMA-ES (Covariance Matrix Adaptation Evolution Strategy) and validated against real data across temporal, distributional, and structural dimensions. Using this validated simulator, we provide three key contributions. First, we show that the calibrated model reproduces key statistical signatures of the empirical data, including activity distributions, post/reshare ratios, and temporal patterns. Second, we apply established misinformation-spreader detection and prevention methods to both empirical and simulated data, progressively removing top-ranked users and showing that the resulting decline in low-quality content is consistent across the two. Third, comparing static (retroactive) and dynamic (in-simulation) moderation across 30 network realisations, we show that static evaluation significantly overestimates the effectiveness of user bans for the most effective detectors: when moderation is applied in real time, compensatory resharing by the remaining users dampens the expected reduction in low-quality content, so static estimates should be read as an upper bound. These findings highlight the necessity of simulation-based evaluation for content moderation policies and contribute a reusable, empirically grounded simulation framework.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-d5be4ed69845d58c", "title": "Physics of Agents: Statistical Mechanics Predicts Collective Behavior of AI Agents", "authors": ["Batu El", "Jinhee Paeng", "Fatih Dinc", "Shiye Su", "Mete Erdogan", "Aneesh Pappu", "Haotian Ye", "Wanjia Zhao", "Surya Ganguli", "James Zou"], "year": 2026, "published": "2026-08-17T13:41:24Z", "updated": "2026-08-17T13:41:24Z", "arxiv_id": "http://arxiv.org/abs/2608.16578v1", "url": "https://arxiv.org/abs/2608.16578v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.MA", "cs.SI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["As agents exchange information and jointly make decisions, their interactions can improve collective reasoning but may also produce herding, polarization, or amplify shared biases.", "On objective questions, communication improves collective accuracy, while on subjective questions it often drifts group opinions toward the right in the political spectrum.", "Given only initial opinions, our model predicts individual trajectories, outperforms all standard baselines, generalizes to unseen community graphs, and reproduces the observed group archetype distributions."], "abstract": "AI agents increasingly operate as part of interacting systems rather than in isolation. As agents exchange information and jointly make decisions, their interactions can improve collective reasoning but may also produce herding, polarization, or amplify shared biases. Understanding and predicting these collective dynamics is therefore important for designing effective and aligned multi-agent systems. Here, we study over 10,000 communities of language-model agents that repeatedly exchange messages and revise their opinions across objective mathematics questions and subjective political statements. Despite substantial diversity in possible behavior, the individual and group dynamics can be represented by three characteristic regimes: indifference, polarization, and consensus. AI agents start indifferent and build conviction as they interact. On objective questions, communication improves collective accuracy, while on subjective questions it often drifts group opinions toward the right in the political spectrum. We explain these observations with a statistical-mechanics formalism in which agents stochastically favor lower social pressure. Given only initial opinions, our model predicts individual trajectories, outperforms all standard baselines, generalizes to unseen community graphs, and reproduces the observed group archetype distributions. Our fitted model parameters reveal the mechanics underlying our key observations: i) communities operate below the critical social temperature, which explains conviction buildup; ii) attractive ties outweigh repulsive ones, which favors consensus; and iii) agents holding the correct answer exert the strongest pull, which drives truth-seeking. Overall, our results demonstrate that collective behavior of AI agents, like that of other complex systems, follows compact and predictive dynamical laws.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-5bf964a7ef5bc23a", "title": "VCE-Skill: Enhancing Skill Self-Evolution with Version-Change Experience", "authors": ["Jianming Chen", "Xuanbin Ye", "Yawen Wang", "Junjie Wang", "Qing Wang", "Fanjiang XU"], "year": 2026, "published": "2026-08-17T13:16:59Z", "updated": "2026-08-17T13:16:59Z", "arxiv_id": "http://arxiv.org/abs/2608.16544v1", "url": "https://arxiv.org/abs/2608.16544v1", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Distillation", "Tool Use"], "benchmarks": [], "models": [], "key_results": ["Extensive experiments demonstrate that VCE-Skill improves skill self-evolution, increasing mean scores by 3.20--4.98 points; transfer experiments further show that the resulting skills achieve stronger cross-model transfer performance."], "abstract": "Agents increasingly rely on reusable skills to encode task knowledge, tool-use procedures, and validation rules. Existing skill self-evolution methods primarily revise skills using execution trajectories collected from current tasks, leaving the evolution knowledge accumulated in public skill version histories largely untapped. Our pilot study reveals a clear complementarity between the two sources: public skill changes provide reusable evolution priors, whereas trajectories provide evidence grounded in the current task. Motivated by this, we propose VCE-Skill, which distills noisy and implementation-specific public skill changes into reusable, structured version-change experience and adaptively fuses it with trajectory-derived proposals from the base evolver, thereby exploiting external experience while retaining task-specific evidence. Extensive experiments demonstrate that VCE-Skill improves skill self-evolution, increasing mean scores by 3.20--4.98 points; transfer experiments further show that the resulting skills achieve stronger cross-model transfer performance. Our work highlights public skill version changes as a previously underexplored yet effective source of prior knowledge and advances trajectory-driven skill self-evolution.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-43944459bf7fcc3e", "title": "JailbreakSkill: Scaling Automated Red-Teaming with Reusable and Ever-Evolving Skills", "authors": ["Xiaoyu Wen", "Jiajia Li", "Zhida He", "Peng Yu", "Chenxu Wang", "Han Qi", "Ziyuan Zhou", "Cheng Jin", "Ying Wen", "Xingcheng Xu", "Shuyue Hu", "Tianhang Zheng", "Chaochao Lu", "Qiaosheng Zhang"], "year": 2026, "published": "2026-08-17T12:03:35Z", "updated": "2026-08-17T12:03:35Z", "arxiv_id": "http://arxiv.org/abs/2608.16465v1", "url": "https://arxiv.org/abs/2608.16465v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": ["GPT-5.4", "yielding"], "key_results": ["Automated red-teaming has produced a growing collection of attack strategies, yet they typically remain scattered across prompts and workflows, making them difficult to systematically integrate, reuse, and improve at scale.", "This evolution lifts macro-average ASR by 17.5 percentage points on AdvBench and 13.4 points on HarmBench, including a 48.6-point gain against GPT-5.4 on AdvBench, while yielding novel attack strategies such as reframing a direct request as an unfinished document-completion task."], "abstract": "Automated red-teaming has produced a growing collection of attack strategies, yet they typically remain scattered across prompts and workflows, making them difficult to systematically integrate, reuse, and improve at scale. We introduce \\textsc{JailbreakSkill}, a skill-centric framework for scaling automated red-teaming through reusable and continuously evolving attack capabilities. \\textsc{JailbreakSkill} packages existing attack strategies into modular, agent-ready skills that can be directly reused and adaptively selected across tasks and target models. Beyond reuse, it closes the loop between attacking and learning: attack experience is used to diagnose, refine, combine, and discover new skills, which are added back to an ever-growing skill library. This evolution lifts macro-average ASR by 17.5 percentage points on AdvBench and 13.4 points on HarmBench, including a 48.6-point gain against GPT-5.4 on AdvBench, while yielding novel attack strategies such as reframing a direct request as an unfinished document-completion task. Several evolved skills also generalize to unseen prompts and target models without further adaptation. Our code is available at https://github.com/BattleWen/JailbreakSkill.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-dc8396240917c738", "title": "MLLM-Guided Semantic Correction for Text-to-Video Generation", "authors": ["Junhao Chen", "Zheqi Lv", "Keting Yin", "Shengyu Zhang", "Zhou Zhao", "Feiyang Chen", "Xinyu Duan", "Baoxing Huai", "Fei Wu"], "year": 2026, "published": "2026-08-17T12:54:10Z", "updated": "2026-08-17T12:54:10Z", "arxiv_id": "http://arxiv.org/abs/2608.16513v1", "url": "https://arxiv.org/abs/2608.16513v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Interpretability", "Video Generation", "VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Our framework achieves diffusion trajectory correction by injecting semantic evaluation signals during video synthesis, enabling the model to optimize the generated content through continuous self-reflection.", "Our method improves semantic alignment, visual fidelity, and temporal consistency without modifying model parameters."], "abstract": "Recent advances in diffusion models and Transformer architectures have led to significant progress in text-to-video generation. However, these models often suffer from semantic errors such as missing objects, incorrect attributes, or mismatched actions. Although some semantic correction methods perform optimization before sampling or refinement after sampling, how to detect and correct semantic deviations during the video generation process remains underexplored. In this paper, we introduce a training-free, interpretable mid-generation correction framework that integrates multimodal large language model (MLLM) feedback directly into the diffusion sampling loop. Our framework achieves diffusion trajectory correction by injecting semantic evaluation signals during video synthesis, enabling the model to optimize the generated content through continuous self-reflection. We propose two key modules: a Semantic Assessment Supervisor that generates intermediate preview frames for semantic evaluations and deviation diagnostics, and a Semantic Modification Assistant that corrects semantic drift during inference via a controllable latent trajectory intervention. Our method improves semantic alignment, visual fidelity, and temporal consistency without modifying model parameters. We validate the effectiveness of our approach through extensive experiments across multiple benchmarks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-2a4bd8908a22ce15", "title": "Remote-Sensing City Layout Extraction with MLLM", "authors": ["Zigan Zhou", "Kai Li", "Yupeng Deng"], "year": 2026, "published": "2026-08-17T12:23:52Z", "updated": "2026-08-17T12:23:52Z", "arxiv_id": "http://arxiv.org/abs/2608.16484v1", "url": "https://arxiv.org/abs/2608.16484v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["phic"], "key_results": ["Remote-sensing systems usually describe urban content with detection boxes, semantic masks, or vector boundaries."], "abstract": "Remote-sensing systems usually describe urban content with detection boxes, semantic masks, or vector boundaries. Such outputs locate classes and support image-plane scoring, yet they do not by themselves constitute an executable layout that retains object identities, typed relations, topology, and regeneration rules. Code-as-City instead casts urban-layout extraction from a single top-down image as constrained code generation with a multimodal large language model (MLLM). An image model first produces an aligned five-class semantic layout prior. Three ordered MLLM passes use the image and this prior to recover roads, land-cover regions and relations, and buildings. Deterministic normalization converts the accumulated records into a city graph and a restricted layout program. Executing the program creates a renderable 3D city layout and an orthographic semantic projection over shared geometry. The projection admits pixel-level comparison with remote-sensing masks, while named objects, relations, and editing operations remain available for synchronized regeneration of both views. Evaluated on the 100 scenes of CityLayout-100, the complete framework obtains 41.1% mean intersection-over-union and 48.3% global intersection-over-union. This result provides quantitative evidence that visual observations can be translated into inspectable, editable city code with coupled planar and 3D outputs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-b0338c441e9494fd", "title": "RISE: Roadside Infrastructure Sequence Understanding across 3D Tracking and Structured Vision-Language Reasoning", "authors": ["Yanbo Jiang", "Haotian Zheng", "Jiahao Wang", "Hanxiao Ren", "Yitao Xu", "Yining Xing", "Zehong Ke", "Hao Cheng", "Yiqian Tu", "Jinhao Li", "Zhiyuan Xuan", "Fang Zhang", "Jianqiang Wang"], "year": 2026, "published": "2026-08-17T12:22:15Z", "updated": "2026-08-17T12:22:15Z", "arxiv_id": "http://arxiv.org/abs/2608.16480v1", "url": "https://arxiv.org/abs/2608.16480v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["On 20 human-reviewed clips from six intersections, the generated tracks achieve 66.9 MOTA within the defined multi-view evaluation scope."], "abstract": "We present RISE (Roadside Infrastructure Sequence Understanding and Evaluation), a framework spanning metric 3D tracking and structured vision-language reasoning in roadside sequences. For metric tracking, our image-only method combines SAM3 video identities with calibration-guided mask agreement for multi-view identity association, recovering persistent 3D tracks without LiDAR or task-specific 3D training. Its calibration-conditioned geometry allows the procedure to be instantiated at different calibrated multi-camera intersections without layout-specific retraining. On 20 human-reviewed clips from six intersections, the generated tracks achieve 66.9 MOTA within the defined multi-view evaluation scope. For structured vision-language reasoning, a human-reviewed MLLM pipeline mines high-value clips and uses a constrained full-context Oracle to construct bbox-grounded predictive QA without exposing future evidence to evaluated models. The resulting RISE-VQA dataset contains 33,910 QA pairs from 557 clips across 16 intersections and 61 roadside views. Its intersection-held-out RISE-Bench evaluates semantic choices, coordinates, future boxes, and interaction sets with deterministic task-specific metrics. Experiments show consistent benefits from domain adaptation and generally from temporal context, while revealing persistent challenges in spatial grounding, future localization, and interaction reasoning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-0013e4357593b76c", "title": "DriveCache: Action-Aware Caching for Driving World Model Inference", "authors": ["Jianchun Yang", "Jian Liang", "Xianda Guo", "Pinhan Fu", "Yanlun Peng", "Conglang Zhang", "Wenke Huang", "Mang Ye"], "year": 2026, "published": "2026-08-17T10:04:49Z", "updated": "2026-08-17T10:04:49Z", "arxiv_id": "http://arxiv.org/abs/2608.16354v1", "url": "https://arxiv.org/abs/2608.16354v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["World Model", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Across three generator configurations, DriveCache improves the overall fidelity-efficiency trade-off over evaluated cache methods."], "abstract": "Driving video generation models support autonomous-driving development by predicting controllable future scenes for simulation, planning evaluation, and offline data generation. Diffusion-based driving generators repeatedly evaluate large backbones across denoising steps, which limits generation throughput. Existing diffusion acceleration methods reduce this cost, but general-purpose designs omit driving signals available before generation, such as ego speed and planned trajectories. Experiments across driving motions show that cache tolerance varies with ego translation and rotation, denoising progress, and consecutive reuse length. We propose DriveCache, a training-free, action-aware controller that uses planned motion to allocate reuse across scenes and dynamic programming to place it across denoising steps under a calibrated response budget. A causal drift check refreshes features and replans the remaining schedule when generation departs from calibration. Across three generator configurations, DriveCache improves the overall fidelity-efficiency trade-off over evaluated cache methods. Our code will be publicly available.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-ce8d1b77c2c05169", "title": "SCALE: State-Calibrated Latent Embeddings for JEPA Planning in the Right Geometry", "authors": ["Jiaming Hu", "Yan Zheng", "Tian Wang"], "year": 2026, "published": "2026-08-17T08:58:55Z", "updated": "2026-08-17T08:58:55Z", "arxiv_id": "http://arxiv.org/abs/2608.16287v1", "url": "https://arxiv.org/abs/2608.16287v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "World Model"], "benchmarks": [], "models": ["yields"], "key_results": ["Joint-embedding predictive world models plan by scoring predicted terminal embeddings against a goal embedding using a cost defined on the representation itself.", "Across five tasks, three planning solvers, and five compute budgets, SCALE improves every task--solver average over LeWM.", "A latent-to-state regression control matches or exceeds SCALE's full-embedding decodability yet leaves latent--state distance alignment essentially unchanged and yields less consistent planning gains."], "abstract": "Joint-embedding predictive world models plan by scoring predicted terminal embeddings against a goal embedding using a cost defined on the representation itself. Two prominent strategies for obtaining non-collapsed representations are to inherit a pretrained feature space, as in DINO-WM, and to learn an embedding end to end with anti-collapse regularization, as in LeWorldModel (LeWM) with SIGReg. These strategies show complementary strengths across tasks. Although task-relevant state is decodable from the full embeddings of both models, DINO-WM's leading principal components usually retain substantially more state information than LeWM's. Because Euclidean planning costs are dominated by high-variance directions, this difference affects how strongly state can influence candidate selection. We propose SCALE (State-CAlibrated Latent Embeddings) to give the end-to-end LeWM representation the favorable geometric property observed in DINO-WM. SCALE induces this property by correlating sampled pairwise latent distances with distances in a standardized task-relevant state space, without replacing LeWM's learned encoder. Across five tasks, three planning solvers, and five compute budgets, SCALE improves every task--solver average over LeWM. A latent-to-state regression control matches or exceeds SCALE's full-embedding decodability yet leaves latent--state distance alignment essentially unchanged and yields less consistent planning gains. SCALE adds a single lightweight training-time regularizer and no planning-time overhead. These results show that planning depends not only on whether task-relevant information is present, but also on whether it shapes the geometry consumed by the planner.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-5f74ff6e5e03b202", "title": "Audio-Visual Segmentation via Depth-Guided Collaborative Modeling", "authors": ["Zhaojin Fu", "Yuyang Hong", "Qi Yang", "Zili Wang", "Kun Ding", "Shiming Xiang", "Bin Fan"], "year": 2026, "published": "2026-08-17T08:55:15Z", "updated": "2026-08-17T08:55:15Z", "arxiv_id": "http://arxiv.org/abs/2608.16285v1", "url": "https://arxiv.org/abs/2608.16285v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["Specifically, we design a Depth-Aware Dynamic Modulator to improve the separation of adjacent objects while preserving intra-object feature consistency.", "Compared to state-of-the-art methods, DGCM-AVS achieves relative improvements of 10.2 percent in M_J and 8.7 percent in M_F on the AVSS dataset."], "abstract": "Audio-Visual Segmentation (AVS) is a fundamental task in multimodal perception that performs pixel-level segmentation of sounding objects in videos by leveraging both visual and audio cues. It has broad applications in video understanding, human-computer interaction, and autonomous driving. However, most existing AVS methods do not explicitly model geometric cues such as relative distance and occlusion, thereby limiting the robustness of cross-modal alignment. In human perception, spatial structure is naturally integrated with audio-visual evidence to accurately localize sounding objects. Motivated by this, we incorporate estimated depth as a spatial structural cue for AVS and propose DGCM-AVS, a tri-modal framework that jointly models audio, visual, and depth information. Specifically, we design a Depth-Aware Dynamic Modulator to improve the separation of adjacent objects while preserving intra-object feature consistency. Furthermore, we propose Depth-Guided Progressive Fusion, which uses depth as an intermediate bridge to progressively align audio cues with visual features. Compared to state-of-the-art methods, DGCM-AVS achieves relative improvements of 10.2 percent in M_J and 8.7 percent in M_F on the AVSS dataset. We believe our study highlights depth as a promising yet underexplored modality for AVS and may encourage further research in this direction.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:37:20.073261+00:00"}
{"id": "evg-a1327781cf590ac8", "title": "Being Kind Isn't Always Being Safe: Diagnosing Affective Hallucination in LLMs", "authors": ["Sewon Kim", "Jiwon Kim", "Seungwoo Shin", "Hyejin Chung", "Daeun Moon", "Yejin Kwon", "Hyunsoo Yoon"], "year": 2025, "published": "2025-08-23T06:55:05Z", "updated": "2026-01-22T05:06:52Z", "arxiv_id": "http://arxiv.org/abs/2508.16921v2", "url": "https://arxiv.org/abs/2508.16921v2", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": ["GPT-4o"], "key_results": ["DPO fine-tuning substantially reduces affective hallucination without compromising reasoning performance, and the Pearson correlation coefficients between GPT-4o and human judgments is also strong (r=0.85) indicating that human evaluations confirm AHaBench as an effective diagnostic tool."], "abstract": "Large Language Models (LLMs) are increasingly engaged in emotionally vulnerable conversations that extend beyond information seeking to moments of personal distress. As they adopt affective tones and simulate empathy, they risk creating the illusion of genuine relational connection. We term this phenomenon Affective Hallucination, referring to emotionally immersive responses that evoke false social presence despite the model's lack of affective capacity. To address this, we introduce AHaBench, a benchmark of 500 mental-health-related prompts with expert-informed reference responses, evaluated along three dimensions: Emotional Enmeshment, Illusion of Presence, and Fostering Overdependence. We further release AHaPairs, a 5K-instance preference dataset enabling Direct Preference Optimization (DPO) for alignment with emotionally responsible behavior. DPO fine-tuning substantially reduces affective hallucination without compromising reasoning performance, and the Pearson correlation coefficients between GPT-4o and human judgments is also strong (r=0.85) indicating that human evaluations confirm AHaBench as an effective diagnostic tool. This work establishes affective hallucination as a distinct safety concern and provides resources for developing LLMs that are both factually reliable and psychologically safe. AHaBench and AHaPairs are accessible via https://huggingface.co/datasets/o0oMiNGo0o/AHaBench, and code for fine-tuning and evaluation are in https://github.com/0oOMiNGOo0/AHaBench. Warning: This paper contains examples of mental health-related language that may be emotionally distressing.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 67174, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["MMLU", "GSM8K", "MATH", "TruthfulQA"], "fulltext_models": ["GPT-4o", "GPT-4o.", "Gpt-4o", "GPT-3.5-turbo", "GPT-5", "LLaMA", "LLaMA3.1-8B-Instruct", "LLaMA3.1-8B", "LLaMA outputs", "LLaMA3"], "matched_methods": ["Preference Optimization", "Safety / Jailbreak"], "checked_on": "2026-08-18T07:48:25.312059+00:00"}}
{"id": "evg-7048f58dd53a5424", "title": "Reinforcement-Guided Hyper-Heuristic Hyperparameter Optimization for Fair and Explainable Spiking Neural Network-Based Financial Fraud Detection", "authors": ["Sadman Mohammad Nasif", "Md Abrar Jahin", "M. F. Mridha"], "year": 2025, "published": "2025-08-23T06:24:15Z", "updated": "2026-03-23T22:34:27Z", "arxiv_id": "http://arxiv.org/abs/2508.16915v3", "url": "https://arxiv.org/abs/2508.16915v3", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Interpretability", "Quant / Trading"], "benchmarks": [], "models": ["phic"], "key_results": ["Evaluated on the Bank Account Fraud (BAF) dataset, the model achieves 90.8% recall at 5% false positive rate, outperforming prior spiking and classical models while maintaining over 98% predictive equality across demographic groups.", "The sparse architecture of CSNPC further reduces energy consumption compared to dense ANNs."], "abstract": "The growing adoption of home banking systems has increased cyberfraud risks, requiring detection models that are accurate, fair, and explainable. While AI methods show promise, they face challenges including computational inefficiency, limited interpretability of spiking neural networks (SNNs), and instability in reinforcement learning (RL)-based hyperparameter optimization. We propose a framework combining a Cortical Spiking Network with Population Coding (CSNPC) and a Reinforcement-Guided Hyper-Heuristic Optimizer (RHOSS). CSNPC leverages population coding for robust classification, while RHOSS applies Q-learning to adaptively select low-level heuristics under fairness and recall constraints. Integrated within the MoSSTI framework, the system incorporates explainable AI via saliency maps and spike activity profiling. Evaluated on the Bank Account Fraud (BAF) dataset, the model achieves 90.8% recall at 5% false positive rate, outperforming prior spiking and classical models while maintaining over 98% predictive equality across demographic groups. Although RHOSS introduces offline optimization cost, it is amortized at deployment. The sparse architecture of CSNPC further reduces energy consumption compared to dense ANNs. Results demonstrate that combining population-coded SNNs with RL-guided hyper-heuristics enables fair, interpretable, and high-performance fraud detection.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 36368, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Interpretability", "Deep Research", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["phic", "phics", "ying"], "matched_methods": ["Interpretability", "Quant / Trading", "RLVR / GRPO"], "checked_on": "2026-08-18T07:48:28.428857+00:00"}}
{"id": "evg-c22ea653b0b9f637", "title": "Unbiased Reasoning for Knowledge-Intensive Tasks in Large Language Models via Conditional Front-Door Adjustment", "authors": ["Bo Zhao", "Yinghao Zhang", "Ziqi Xu", "Yongli Ren", "Xiuzhen Zhang", "Renqiang Luo", "Zaiwen Feng", "Feng Xia"], "year": 2025, "published": "2025-08-23T05:52:39Z", "updated": "2025-08-23T05:52:39Z", "arxiv_id": "http://arxiv.org/abs/2508.16910v1", "url": "https://arxiv.org/abs/2508.16910v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Memory / RAG"], "benchmarks": [], "models": ["ying"], "key_results": ["Extensive experiments across multiple LLMs and benchmark datasets demonstrate that CFD-Prompting significantly outperforms existing baselines in both accuracy and robustness."], "abstract": "Large Language Models (LLMs) have shown impressive capabilities in natural language processing but still struggle to perform well on knowledge-intensive tasks that require deep reasoning and the integration of external knowledge. Although methods such as Retrieval-Augmented Generation (RAG) and Chain-of-Thought (CoT) have been proposed to enhance LLMs with external knowledge, they still suffer from internal bias in LLMs, which often leads to incorrect answers. In this paper, we propose a novel causal prompting framework, Conditional Front-Door Prompting (CFD-Prompting), which enables the unbiased estimation of the causal effect between the query and the answer, conditional on external knowledge, while mitigating internal bias. By constructing counterfactual external knowledge, our framework simulates how the query behaves under varying contexts, addressing the challenge that the query is fixed and is not amenable to direct causal intervention. Compared to the standard front-door adjustment, the conditional variant operates under weaker assumptions, enhancing both robustness and generalisability of the reasoning process. Extensive experiments across multiple LLMs and benchmark datasets demonstrate that CFD-Prompting significantly outperforms existing baselines in both accuracy and robustness.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 35382, "fulltext_methods": ["Chain-of-Thought", "KV Cache", "Memory / RAG"], "fulltext_benchmarks": ["DROP", "HotpotQA"], "fulltext_models": ["GPT-3.5", "LLaMA-2", "LLaMA-3", "Llama-2-7b-chat-hf", "Llama-3-8B-Instruct", "LLaMA-2.", "LLaMA-3.", "phical", "Yinghao", "ying"], "matched_methods": ["Chain-of-Thought", "Memory / RAG"], "checked_on": "2026-08-18T07:48:30.640977+00:00"}}
{"id": "evg-964669cd2625570d", "title": "Dream to Chat: Model-based Reinforcement Learning on Dialogues with User Belief Modeling", "authors": ["Yue Zhao", "Xiaoyu Wang", "Dan Wang", "Zhonglin Jiang", "Qingqing Gu", "Teng Chen", "Ningyuan Xi", "Jinxian Qu", "Yong Chen", "Luo Ji"], "year": 2025, "published": "2025-08-23T02:24:03Z", "updated": "2025-09-26T02:30:43Z", "arxiv_id": "http://arxiv.org/abs/2508.16876v3", "url": "https://arxiv.org/abs/2508.16876v3", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "World Model"], "benchmarks": [], "models": [], "key_results": ["Experiments show that the pretrained dialogue world model can achieve state-of-the-art performances on emotion classification and sentiment identification, while dialogue quality is also enhanced by joint training of the policy, critic and dialogue world model."], "abstract": "World models have been widely utilized in robotics, gaming, and auto-driving. However, their applications on natural language tasks are relatively limited. In this paper, we construct the dialogue world model, which could predict the user's emotion, sentiment, and intention, and future utterances. By defining a POMDP, we argue emotion, sentiment and intention can be modeled as the user belief and solved by maximizing the information bottleneck. By this user belief modeling, we apply the model-based reinforcement learning framework to the dialogue system, and propose a framework called DreamCUB. Experiments show that the pretrained dialogue world model can achieve state-of-the-art performances on emotion classification and sentiment identification, while dialogue quality is also enhanced by joint training of the policy, critic and dialogue world model. Further analysis shows that this manner holds a reasonable exploration-exploitation balance and also transfers well to out-of-domain scenarios such as empathetic dialogues.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 16213, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Preference Optimization", "World Model"], "fulltext_benchmarks": [], "fulltext_models": ["llama2-7b-chat", "llama-chat-7b", "llama2-13b-chat", "llama-chat-13b", "llama3-8b-instruct", "Llama3.1-8B-Instruct", "LLama.", "Llama3-based", "ying", "yields"], "matched_methods": ["RLVR / GRPO", "World Model"], "checked_on": "2026-08-18T07:48:35.987013+00:00"}}
{"id": "evg-f469e87d0194dc45", "title": "Learning from Diverse Reasoning Paths with Routing and Collaboration", "authors": ["Zhenyu Lei", "Zhen Tan", "Song Wang", "Yaochen Zhu", "Zihan Chen", "Yushun Dong", "Jundong Li"], "year": 2025, "published": "2025-08-23T01:15:57Z", "updated": "2025-08-23T01:15:57Z", "arxiv_id": "http://arxiv.org/abs/2508.16861v1", "url": "https://arxiv.org/abs/2508.16861v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["Advances in large language models (LLMs) significantly enhance reasoning capabilities but their deployment is restricted in resource-constrained scenarios."], "abstract": "Advances in large language models (LLMs) significantly enhance reasoning capabilities but their deployment is restricted in resource-constrained scenarios. Knowledge distillation addresses this by transferring knowledge from powerful teacher models to compact and transparent students. However, effectively capturing the teacher's comprehensive reasoning is challenging due to conventional token-level supervision's limited scope. Using multiple reasoning paths per query alleviates this problem, but treating each path identically is suboptimal as paths vary widely in quality and suitability across tasks and models. We propose Quality-filtered Routing with Cooperative Distillation (QR-Distill), combining path quality filtering, conditional routing, and cooperative peer teaching. First, quality filtering retains only correct reasoning paths scored by an LLM-based evaluation. Second, conditional routing dynamically assigns paths tailored to each student's current learning state. Finally, cooperative peer teaching enables students to mutually distill diverse insights, addressing knowledge gaps and biases toward specific reasoning styles. Experiments demonstrate QR-Distill's superiority over traditional single- and multi-path distillation methods. Ablation studies further highlight the importance of each component including quality filtering, conditional routing, and peer teaching in effective knowledge transfer. Our code is available at https://github.com/LzyFischer/Distill.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 32462, "fulltext_methods": ["Chain-of-Thought", "Test-time Scaling", "Distillation", "Preference Optimization", "Safety / Jailbreak", "Multi-Agent", "Memory / RAG"], "fulltext_benchmarks": ["GSM8K", "MATH"], "fulltext_models": ["Gemini-1.5-Pro-001", "Gemini-1.5-Pro-001 Team et al. 2024a as the black-box teacher model", "Gemini-1.5 model. Although Gemini is a strong teacher", "Mistral-7B-Instruct Zero-shot Kojima et al. 2022 53.89 73.68 10.42 43.92 39.64 44.31 SKD Li et al. 2023b 63.76 74.66 12.48 44.90 48.50 48.86 Distill Step-by-Step Hsieh et al. 2023 64.19 75.32 11.54 44.42 49.63 49.02 Rephrase Question Yu et al. 2024 65.07 74.51 12.98 43.58 45.51 48.33 Question Aug Li et al. 2024c 65.07 73.32 13.64 42.20 47.21 48.29 Answer Aug Yu et al. 2024 66.38 76.77 14.78 45.01 49.12 50.41 RevTHINK Chen et al. 2024a 70.97 78.50 15.28 48.58 70.40 56.75 QR-Distill", "Mistral-7B-Instruct and Gemma-7B-Instruct. Results are reported from prior work unless noted. Best values are bolded. 3.1 Backbone Models We use Gemini-1.5-Pro-001 Team et al. 2024a as the black-box teacher model", "Mistral-7B-Instruct-v0.3 Jiang et al. 2024 and Gemma-7B-Instruct Team et al. 2024b", "Mistral and for Gemma", "Mistral and with Gemma", "Mistral across most datasets. Interestingly", "Mistral under QR-Distill"], "matched_methods": ["Distillation"], "checked_on": "2026-08-18T07:48:42.165098+00:00"}}
{"id": "evg-241a8c0c5dadc91c", "title": "RADAR: A Reasoning-Guided Attribution Framework for Explainable Visual Data Analysis", "authors": ["Anku Rani", "Aparna Garimella", "Apoorv Saxena", "Balaji Vasan Srinivasan", "Paul Pu Liang"], "year": 2025, "published": "2025-08-23T00:42:43Z", "updated": "2025-08-23T00:42:43Z", "arxiv_id": "http://arxiv.org/abs/2508.16850v1", "url": "https://arxiv.org/abs/2508.16850v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization", "Interpretability", "VLM"], "benchmarks": [], "models": ["BERT"], "key_results": ["Experimental results demonstrate that our reasoning-guided approach improves attribution accuracy by 15% compared to baseline methods, and enhanced attribution capabilities translate to stronger answer generation, achieving an average BERTScore of $\\sim$ 0.90, indicating high alignment with ground truth responses."], "abstract": "Data visualizations like charts are fundamental tools for quantitative analysis and decision-making across fields, requiring accurate interpretation and mathematical reasoning. The emergence of Multimodal Large Language Models (MLLMs) offers promising capabilities for automated visual data analysis, such as processing charts, answering questions, and generating summaries. However, they provide no visibility into which parts of the visual data informed their conclusions; this black-box nature poses significant challenges to real-world trust and adoption. In this paper, we take the first major step towards evaluating and enhancing the capabilities of MLLMs to attribute their reasoning process by highlighting the specific regions in charts and graphs that justify model answers. To this end, we contribute RADAR, a semi-automatic approach to obtain a benchmark dataset comprising 17,819 diverse samples with charts, questions, reasoning steps, and attribution annotations. We also introduce a method that provides attribution for chart-based mathematical reasoning. Experimental results demonstrate that our reasoning-guided approach improves attribution accuracy by 15% compared to baseline methods, and enhanced attribution capabilities translate to stronger answer generation, achieving an average BERTScore of $\\sim$ 0.90, indicating high alignment with ground truth responses. This advancement represents a significant step toward more interpretable and trustworthy chart analysis systems, enabling users to verify and understand model decisions through reasoning and attribution.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 50822, "fulltext_methods": ["Chain-of-Thought", "Preference Optimization", "Interpretability", "Tool Use", "Memory / RAG", "Computer Use", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["MATH", "ChartQA"], "fulltext_models": ["GPT-4o", "GPT-4v", "Gpt-4o", "Gpt-4v", "GPT-4o44", "GPT-4v55", "gpt-4-vision-preview", "Claude 3.5 Sonnet and GPT-4o as our primary models for analysis. We also include GPT-4v for its vision capabilities. To assess the performance on reasoning generation", "Claude-3.5-sonnet 96", "Claude 3.5 Sonnet demonstrates strong answer generation"], "matched_methods": ["Interpretability", "Preference Optimization", "VLM"], "checked_on": "2026-08-18T07:48:45.691291+00:00"}}
{"id": "evg-3015d852ff7f6b12", "title": "NinA: Normalizing Flows in Action. Training VLA Models with Normalizing Flows", "authors": ["Denis Tarasov", "Alexander Nikulin", "Ilya Zisman", "Albina Klepach", "Nikita Lyubaykin", "Andrei Polubarov", "Alexander Derevyagin", "Vladislav Kurenkov"], "year": 2025, "published": "2025-08-23T00:02:15Z", "updated": "2025-10-14T10:06:39Z", "arxiv_id": "http://arxiv.org/abs/2508.16845v2", "url": "https://arxiv.org/abs/2508.16845v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Video Generation", "VLM"], "benchmarks": [], "models": [], "key_results": ["Our experiments show that NinA matches the performance of its diffusion-based counterpart under the same training regime, while achieving substantially faster inference."], "abstract": "Recent advances in Vision-Language-Action (VLA) models have established a two-component architecture, where a pre-trained Vision-Language Model (VLM) encodes visual observations and task descriptions, and an action decoder maps these representations to continuous actions. Diffusion models have been widely adopted as action decoders due to their ability to model complex, multimodal action distributions. However, they require multiple iterative denoising steps at inference time or downstream techniques to speed up sampling, limiting their practicality in real-world settings where high-frequency control is crucial. In this work, we present NinA (Normalizing Flows in Action), a fast and expressive alternative to diffusion-based decoders for VLAs. NinA replaces the diffusion action decoder with a Normalizing Flow (NF) that enables one-shot sampling through an invertible transformation, significantly reducing inference time. We integrate NinA into the FLOWER VLA architecture and fine-tune on the LIBERO benchmark. Our experiments show that NinA matches the performance of its diffusion-based counterpart under the same training regime, while achieving substantially faster inference. These results suggest that NinA offers a promising path toward efficient, high-frequency VLA control without compromising performance.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 22034, "fulltext_methods": ["RLVR / GRPO", "Test-time Scaling", "Distillation", "Interpretability", "Memory / RAG", "Deep Research", "Video Generation", "VLM"], "fulltext_benchmarks": ["DROP"], "fulltext_models": ["ying", "Yifeng", "Yihao", "yi", "Bert"], "matched_methods": ["VLM", "Video Generation"], "checked_on": "2026-08-18T07:49:00.437254+00:00"}}
{"id": "evg-9854679e8158f1e5", "title": "If We May De-Presuppose: Robustly Verifying Claims through Presupposition-Free Question Decomposition", "authors": ["Shubhashis Roy Dipta", "Francis Ferraro"], "year": 2025, "published": "2025-08-22T23:34:24Z", "updated": "2025-09-27T02:54:30Z", "arxiv_id": "http://arxiv.org/abs/2508.16838v2", "url": "https://arxiv.org/abs/2508.16838v2", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Extensive experiments across multiple prompts, datasets, and LLMs reveal that even state-of-the-art models remain susceptible to prompt variance and presupposition.", "Our method consistently mitigates these issues, achieving up to a 2-5% improvement."], "abstract": "Prior work has shown that presupposition in generated questions can introduce unverified assumptions, leading to inconsistencies in claim verification. Additionally, prompt sensitivity remains a significant challenge for large language models (LLMs), resulting in performance variance as high as 3-6%. While recent advancements have reduced this gap, our study demonstrates that prompt sensitivity remains a persistent issue. To address this, we propose a structured and robust claim verification framework that reasons through presupposition-free, decomposed questions. Extensive experiments across multiple prompts, datasets, and LLMs reveal that even state-of-the-art models remain susceptible to prompt variance and presupposition. Our method consistently mitigates these issues, achieving up to a 2-5% improvement.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 41305, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Preference Optimization", "Memory / RAG", "Video Generation", "VLM", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["Llama 3", "Llama 2", "Qwen", "Qwen3-32B", "Qwen2.5", "DeepSeek-AI", "DeepSeek-R1", "R1", "Philippe", "ying"], "matched_methods": [], "checked_on": "2026-08-18T07:49:06.705696+00:00"}}
{"id": "evg-f945ded0c59d6837", "title": "PuzzleJAX: A Benchmark for Reasoning and Learning", "authors": ["Sam Earle", "Graham Todd", "Yuchen Li", "Ahmed Khalifa", "Muhammad Umair Nasir", "Zehua Jiang", "Andrzej Banburski-Fahey", "Julian Togelius"], "year": 2025, "published": "2025-08-22T22:40:58Z", "updated": "2025-08-22T22:40:58Z", "arxiv_id": "http://arxiv.org/abs/2508.16821v1", "url": "https://arxiv.org/abs/2508.16821v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Search / MCTS"], "benchmarks": [], "models": [], "key_results": ["We introduce PuzzleJAX, a GPU-accelerated puzzle game engine and description language designed to support rapid benchmarking of tree search, reinforcement learning, and LLM reasoning abilities."], "abstract": "We introduce PuzzleJAX, a GPU-accelerated puzzle game engine and description language designed to support rapid benchmarking of tree search, reinforcement learning, and LLM reasoning abilities. Unlike existing GPU-accelerated learning environments that provide hard-coded implementations of fixed sets of games, PuzzleJAX allows dynamic compilation of any game expressible in its domain-specific language (DSL). This DSL follows PuzzleScript, which is a popular and accessible online game engine for designing puzzle games. In this paper, we validate in PuzzleJAX several hundred of the thousands of games designed in PuzzleScript by both professional designers and casual creators since its release in 2013, thereby demonstrating PuzzleJAX's coverage of an expansive, expressive, and human-relevant space of tasks. By analyzing the performance of search, learning, and language models on these games, we show that PuzzleJAX can naturally express tasks that are both simple and intuitive to understand, yet often deeply challenging to master, requiring a combination of control, planning, and high-level insight.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 64921, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Long Context", "Preference Optimization", "Interpretability", "Multi-Agent", "Memory / RAG", "VLM", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-4o-mini", "o1995temporal", "o3-mini", "Gemini showed limited success across most tasks", "Gemini", "Qwen-plus", "Deepseek-chat.", "DeepSeek-R1", "R1", "phisticated"], "matched_methods": ["RLVR / GRPO", "Search / MCTS"], "checked_on": "2026-08-18T07:49:10.092859+00:00"}}
{"id": "evg-28ba92181d8faeac", "title": "Compiling Prompts, Not Crafting Them: A Reproducible Workflow for AI-Assisted Evidence Synthesis", "authors": ["Teo Susnjak"], "year": 2025, "published": "2025-08-22T21:37:49Z", "updated": "2025-08-22T21:37:49Z", "arxiv_id": "http://arxiv.org/abs/2509.00038v1", "url": "https://arxiv.org/abs/2509.00038v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Deep Research"], "benchmarks": [], "models": [], "key_results": ["Large language models (LLMs) offer significant potential to accelerate systematic literature reviews (SLRs), yet current approaches often rely on brittle, manually crafted prompts that compromise reliability and reproducibility."], "abstract": "Large language models (LLMs) offer significant potential to accelerate systematic literature reviews (SLRs), yet current approaches often rely on brittle, manually crafted prompts that compromise reliability and reproducibility. This fragility undermines scientific confidence in LLM-assisted evidence synthesis. In response, this work adapts recent advances in declarative prompt optimisation, developed for general-purpose LLM applications, and demonstrates their applicability to the domain of SLR automation. This research proposes a structured, domain-specific framework that embeds task declarations, test suites, and automated prompt tuning into a reproducible SLR workflow. These emerging methods are translated into a concrete blueprint with working code examples, enabling researchers to construct verifiable LLM pipelines that align with established principles of transparency and rigour in evidence synthesis. This is a novel application of such approaches to SLR pipelines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": false, "verification": {"source": "none", "status": "fetch-error", "error": "fulltext fetch failed after 3 attempts: <urlopen error Tunnel connection failed: 503 Service Unavailable>", "checked_on": "2026-08-18T07:49:26.464110+00:00"}}
{"id": "evg-6747a4c3febaaa9e", "title": "Autonomous UAV Flight Navigation in Confined Spaces: A Reinforcement Learning Approach", "authors": ["Marco S. Tayar", "Lucas K. de Oliveira", "Felipe Andrade G. Tommaselli", "Juliano D. Negri", "Thiago H. Segreto", "Ricardo V. Godoy", "Marcelo Becker"], "year": 2025, "published": "2025-08-22T21:29:59Z", "updated": "2025-10-11T16:30:57Z", "arxiv_id": "http://arxiv.org/abs/2508.16807v2", "url": "https://arxiv.org/abs/2508.16807v2", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.LG", "eess.SY"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["This paper directly investigates this trade-off by comparing a leading on-policy algorithm, Proximal Policy Optimization (PPO), against an off-policy counterpart, Soft Actor-Critic (SAC), for precision flight in procedurally generated ducts within a high-fidelity simulator."], "abstract": "Autonomous UAV inspection of confined industrial infrastructure, such as ventilation ducts, demands robust navigation policies where collisions are unacceptable. While Deep Reinforcement Learning (DRL) offers a powerful paradigm for developing such policies, it presents a critical trade-off between on-policy and off-policy algorithms. Off-policy methods promise high sample efficiency, a vital trait for minimizing costly and unsafe real-world fine-tuning. In contrast, on-policy methods often exhibit greater training stability, which is essential for reliable convergence in hazard-dense environments. This paper directly investigates this trade-off by comparing a leading on-policy algorithm, Proximal Policy Optimization (PPO), against an off-policy counterpart, Soft Actor-Critic (SAC), for precision flight in procedurally generated ducts within a high-fidelity simulator. Our results show that PPO consistently learned a stable, collision-free policy that completed the entire course. In contrast, SAC failed to find a complete solution, converging to a suboptimal policy that navigated only the initial segments before failure. This work provides evidence that for high-precision, safety-critical navigation tasks, the reliable convergence of a well-established on-policy method can be more decisive than the nominal sample efficiency of an off-policy algorithm.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 27245, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Deep Research", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["phic", "ying"], "matched_methods": ["Preference Optimization", "RLVR / GRPO"], "checked_on": "2026-08-18T07:49:33.500863+00:00"}}
{"id": "evg-5fecf5a0e61d79e2", "title": "Assess and Prompt: A Generative RL Framework for Improving Engagement in Online Mental Health Communities", "authors": ["Bhagesh Gaur", "Karan Gupta", "Aseem Srivastava", "Manish Gupta", "Md Shad Akhtar"], "year": 2025, "published": "2025-08-22T20:40:05Z", "updated": "2025-08-22T20:40:05Z", "arxiv_id": "http://arxiv.org/abs/2508.16788v1", "url": "https://arxiv.org/abs/2508.16788v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Verifier / PRM"], "benchmarks": [], "models": [], "key_results": ["Empirical results across four notable language models demonstrate significant improvements in attribute elicitation and user engagement."], "abstract": "Online Mental Health Communities (OMHCs) provide crucial peer and expert support, yet many posts remain unanswered due to missing support attributes that signal the need for help. We present a novel framework that identifies these gaps and prompts users to enrich their posts, thereby improving engagement. To support this, we introduce REDDME, a new dataset of 4,760 posts from mental health subreddits annotated for the span and intensity of three key support attributes: event what happened?, effect what did the user experience?, and requirement what support they need?. Next, we devise a hierarchical taxonomy, CueTaxo, of support attributes for controlled question generation. Further, we propose MH-COPILOT, a reinforcement learning-based system that integrates (a) contextual attribute-span identification, (b) support attribute intensity classification, (c) controlled question generation via a hierarchical taxonomy, and (d) a verifier for reward modeling. Our model dynamically assesses posts for the presence/absence of support attributes, and generates targeted prompts to elicit missing information. Empirical results across four notable language models demonstrate significant improvements in attribute elicitation and user engagement. A human evaluation further validates the model's effectiveness in real-world OMHC settings.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 74541, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Quantization", "Long Context", "Preference Optimization", "Interpretability", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["GPT-4", "Llama-3", "Llama-3-70b", "llama 3", "LLaMA 3-instruct", "R1", "Mistral", "Mistral Jiang et al. 2023 Zero Shot 49.15 37.47 47.58 50.69 45.52 40.95 35.68 87.01 89.77 88.36 61.63 Phi-3 Abdin et al. 2024 39.22 26.98 36.29 43.80 40.04 37.48 34.46 85.28 89.47 87.32 50.09 Llama-3 Grattafiori et al. 2024 43.60 28.00 40.80 50.10 45.30 42.20 38.61 87.00 89.12 88.00 48.50 Gemma-2 Riviere et al. 2024 45.24 31.64 42.03 52.22 48.23 45.48 42.41 87.27 89.96 88.58 61.51 Mistral Jiang et al. 2023 SFT 72.39 62.25 69.71 82.25 80.00 78.12 76.21 96.55 95.58 96.04 79.83 Phi-3 Abdin et al. 2024 66.88 56.66 64.14 78.97 76.48 74.57 72.69 96.07 94.53 95.25 75.65 Llama-3 Grattafiori et al. 2024 71.30 61.30 68.40 82.30 79.70 77.82 75.91 96.30 95.30 95.80 79.00 Gemma-2 Riviere et al. 2024 68.20 58.04 65.58 80.39 77.87 75.95 74.00 96.10 94.94 95.48 76.98 MH-Copilot SFT", "Mistral yield competing scores. However", "MistralCueTaxo yields performance comparable to MH-Copilot. Later"], "matched_methods": ["RLVR / GRPO", "Verifier / PRM"], "checked_on": "2026-08-18T07:49:37.505092+00:00"}}
{"id": "evg-ec8c48021828b0d4", "title": "Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling", "authors": ["Ivan Rodkin", "Daniil Orel", "Konstantin Smirnov", "Arman Bolatov", "Bilal Elbouardi", "Besher Hassan", "Yuri Kuratov", "Aydar Bulatov", "Preslav Nakov", "Timothy Baldwin", "Artem Shelmanov", "Mikhail Burtsev"], "year": 2025, "published": "2025-08-22T18:57:08Z", "updated": "2026-05-07T14:29:44Z", "arxiv_id": "http://arxiv.org/abs/2508.16745v3", "url": "https://arxiv.org/abs/2508.16745v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Test-time Scaling", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["We find that most neural architectures trained from scratch can learn rule inference and achieve high next-step accuracy, but performance drops sharply as the required number of intermediate reasoning steps increases.", "Experiments show that increasing model depth is crucial, and extending effective depth via recurrence, memory, or test-time compute improves results but remains bounded."], "abstract": "Reasoning is a core capability of large language models, yet how multi-step reasoning is learned and executed remains unclear. We study this question in a controlled cellular-automata (1dCA) framework that excludes memorisation by using disjoint training and test rules. Given a short state sequence, the model is required to infer the hidden local rule and then chain it to predict multiple future steps. Our evaluation shows that LLMs largely fail to reliably solve a natural-language proxy of the proposed task. We find that most neural architectures trained from scratch can learn rule inference and achieve high next-step accuracy, but performance drops sharply as the required number of intermediate reasoning steps increases. Experiments show that increasing model depth is crucial, and extending effective depth via recurrence, memory, or test-time compute improves results but remains bounded. The code is available on github: https://github.com/RodkinIvan/associative-recurrent-memory-transformer/tree/ACT", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 69566, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Distillation", "Long Context", "Preference Optimization", "Interpretability", "Memory / RAG"], "fulltext_benchmarks": ["MATH", "AIME"], "fulltext_models": ["o1", "gemini-with-deep-think-officially-achieves-gold-medal-standard-at-the-international-mathematical-olympiad", "DeepSeek", "Deepseek-r1", "Deepseekmath", "R1", "r1", "ying", "yields", "yield"], "matched_methods": ["Memory / RAG", "Test-time Scaling"], "checked_on": "2026-08-18T07:49:40.984802+00:00"}}
{"id": "evg-a5fe3a983bf047c9", "title": "WST: Weak-to-Strong Knowledge Transfer via Reinforcement Learning", "authors": ["Haosen Ge", "Shuo Li", "Lianghuan Huang"], "year": 2025, "published": "2025-08-22T18:33:06Z", "updated": "2025-08-22T18:33:06Z", "arxiv_id": "http://arxiv.org/abs/2508.16741v1", "url": "https://arxiv.org/abs/2508.16741v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Distillation", "Preference Optimization"], "benchmarks": ["GSM8K", "MATH"], "models": ["GPT-4o-mini", "Llama-70B.", "yielding"], "key_results": ["Using reinforcement learning, the Teacher Model's instructions are iteratively improved based on the Student Model's outcomes, yielding substantial gains across reasoning (MATH-500, GSM8K) and alignment (HH-RLHF) benchmarks - 98% on MATH-500 and 134% on HH-RLHF - and surpassing baselines such as GPT-4o-mini and Llama-70B."], "abstract": "Effective prompt engineering remains a challenging task for many applications. We introduce Weak-to-Strong Transfer (WST), an automatic prompt engineering framework where a small \"Teacher\" model generates instructions that enhance the performance of a much larger \"Student\" model. Unlike prior work, WST requires only a weak teacher, making it efficient and broadly applicable in settings where large models are closed-source or difficult to fine-tune. Using reinforcement learning, the Teacher Model's instructions are iteratively improved based on the Student Model's outcomes, yielding substantial gains across reasoning (MATH-500, GSM8K) and alignment (HH-RLHF) benchmarks - 98% on MATH-500 and 134% on HH-RLHF - and surpassing baselines such as GPT-4o-mini and Llama-70B. These results demonstrate that small models can reliably scaffold larger ones, unlocking latent capabilities while avoiding misleading prompts that stronger teachers may introduce, establishing WST as a scalable solution for efficient and safe LLM prompt refinement.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 13356, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Chain-of-Thought", "Distillation", "Preference Optimization", "Safety / Jailbreak"], "fulltext_benchmarks": ["GSM8K", "MATH"], "fulltext_models": ["GPT-4o-mini", "GPT-4o-mini.", "GPT-4o", "o1", "Llama-70B.", "Llama 70B", "Llama-3-8B-Instruct", "Llama-3.3-70B-Instruct", "Qwen2.5-Math-1.5B-Instruct", "Qwen2.5-0.5B-Instruct"], "matched_methods": ["Distillation", "Preference Optimization", "RLVR / GRPO"], "checked_on": "2026-08-18T07:49:43.880430+00:00"}}
{"id": "evg-2383d25a926cadbc", "title": "Error Reflection Prompting: Can Large Language Models Successfully Understand Errors?", "authors": ["Jason Li", "Lauren Yraola", "Kevin Zhu", "Sean O'Brien"], "year": 2025, "published": "2025-08-22T18:02:36Z", "updated": "2025-08-22T18:02:36Z", "arxiv_id": "http://arxiv.org/abs/2508.16729v1", "url": "https://arxiv.org/abs/2508.16729v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Interpretability"], "benchmarks": [], "models": [], "key_results": ["Prompting methods for language models, such as Chain-of-thought (CoT), present intuitive step-by-step processes for problem solving."], "abstract": "Prompting methods for language models, such as Chain-of-thought (CoT), present intuitive step-by-step processes for problem solving. These methodologies aim to equip models with a better understanding of the correct procedures for addressing a given task. Despite these advancements, CoT lacks the ability of reflection and error correction, potentially causing a model to perpetuate mistakes and errors. Therefore, inspired by the human ability for said tasks, we propose Error Reflection Prompting (ERP) to further enhance reasoning in language models. Building upon CoT, ERP is a method comprised of an incorrect answer, error recognition, and a correct answer. This process enables the model to recognize types of errors and the steps that lead to incorrect answers, allowing the model to better discern which steps to avoid and which to take. The model is able to generate the error outlines itself with automated ERP generation, allowing for error recognition and correction to be integrated into the reasoning chain and produce scalability and reliability in the process. The results demonstrate that ERP serves as a versatile supplement to conventional CoT, ultimately contributing to more robust and capable reasoning abilities along with increased interpretability in how models ultimately reach their errors.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 35859, "fulltext_methods": ["Verifier / PRM", "Chain-of-Thought", "Interpretability"], "fulltext_benchmarks": ["GSM8K", "MATH"], "fulltext_models": ["GPT-4", "gpt-4-1106-preview", "Gpt-4", "GPT-3.5", "gpt-3.5-turbo-0613", "gpt-3", "phisticated", "Phil", "ying", "yielded"], "matched_methods": ["Chain-of-Thought", "Interpretability"], "checked_on": "2026-08-18T07:49:55.203560+00:00"}}
{"id": "evg-68ddef1dda3c6549", "title": "Hierarchical Decision-Making for Autonomous Navigation: Integrating Deep Reinforcement Learning and Fuzzy Logic in Four-Wheel Independent Steering and Driving Systems", "authors": ["Yizhi Wang", "Degang Xu", "Yongfang Xie", "Shuzhong Tan", "Xianan Zhou", "Peng Chen"], "year": 2025, "published": "2025-08-22T17:57:56Z", "updated": "2025-08-22T17:57:56Z", "arxiv_id": "http://arxiv.org/abs/2508.16574v1", "url": "https://arxiv.org/abs/2508.16574v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["ying"], "key_results": ["Simulation experiments demonstrate that the proposed framework outperforms traditional navigation methods, offering enhanced training efficiency and stability and mitigating erratic behaviors compared to purely DRL-based solutions."], "abstract": "This paper presents a hierarchical decision-making framework for autonomous navigation in four-wheel independent steering and driving (4WISD) systems. The proposed approach integrates deep reinforcement learning (DRL) for high-level navigation with fuzzy logic for low-level control to ensure both task performance and physical feasibility. The DRL agent generates global motion commands, while the fuzzy logic controller enforces kinematic constraints to prevent mechanical strain and wheel slippage. Simulation experiments demonstrate that the proposed framework outperforms traditional navigation methods, offering enhanced training efficiency and stability and mitigating erratic behaviors compared to purely DRL-based solutions. Real-world validations further confirm the framework's ability to navigate safely and effectively in dynamic industrial settings. Overall, this work provides a scalable and reliable solution for deploying 4WISD mobile robots in complex, real-world scenarios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 28371, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Safety / Jailbreak"], "fulltext_benchmarks": [], "fulltext_models": ["Yizhi", "ying", "Yilmaz"], "matched_methods": ["RLVR / GRPO"], "checked_on": "2026-08-18T07:49:57.522469+00:00"}}
{"id": "evg-586cf3e60ab32f23", "title": "Murakkab: Resource-Efficient Agentic Workflow Orchestration in Cloud Platforms", "authors": ["Gohar Irfan Chaudhry", "Esha Choukse", "Haoran Qiu", "Íñigo Goiri", "Rodrigo Fonseca", "Adam Belay", "Ricardo Bianchini"], "year": 2025, "published": "2025-08-22T17:41:27Z", "updated": "2025-09-03T16:28:25Z", "arxiv_id": "http://arxiv.org/abs/2508.18298v2", "url": "https://arxiv.org/abs/2508.18298v2", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.AI", "cs.SE"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["By exposing the internal structure of agentic workflows, Murakkab enables cross-layer optimization that existing frameworks and cloud schedulers cannot achieve.", "Our evaluation on diverse workflows shows that Murakkab reduces GPU usage by up to 2.8$\\times$, energy consumption by 3.7$\\times$, and cost by 4.3$\\times$ while maintaining SLOs."], "abstract": "Agentic workflows commonly coordinate multiple models and tools with complex control logic. They are quickly becoming the dominant paradigm for AI applications. However, serving them remains inefficient with today's frameworks. The key problem is that they expose workflows as opaque sequences of model and tool calls that tightly couple agent logic with model and hardware choices. Often, these workflow components are fragmented across different entities, preventing systems from reasoning about trade-offs across accuracy, latency, energy, and cost. This leads to resource waste and degraded service-level objectives (SLOs).\n  We present Murakkab, a resource-efficient serving system for agentic workflows. Murakkab introduces a declarative abstraction that decouples workflow specification from execution configuration. A profile-guided optimizer and adaptive runtime jointly manage the full stack: orchestrating workflow components, mapping them to models and hardware, and dynamically reconfiguring execution to satisfy user-defined SLOs. By exposing the internal structure of agentic workflows, Murakkab enables cross-layer optimization that existing frameworks and cloud schedulers cannot achieve.\n  Our evaluation on diverse workflows shows that Murakkab reduces GPU usage by up to 2.8$\\times$, energy consumption by 3.7$\\times$, and cost by 4.3$\\times$ while maintaining SLOs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 79111, "fulltext_methods": ["RLVR / GRPO", "Test-time Scaling", "KV Cache", "Multi-Agent", "Memory / RAG", "Deep Research", "VLM", "Quant / Trading"], "fulltext_benchmarks": ["MATH", "HumanEval", "Video-MME", "DROP"], "fulltext_models": ["o3-and-o4-mini", "Claude Code", "claude-code. Accessed", "LlamaIndex", "Llama-3.2", "Llama 3", "LlamaIndex.", "llama_index.", "Llama", "Qwen-32B"], "matched_methods": ["Multi-Agent"], "checked_on": "2026-08-18T07:50:07.272020+00:00"}}
{"id": "evg-b05b770e73761f83", "title": "TinyML Towards Industry 4.0: Resource-Efficient Process Monitoring of a Milling Machine", "authors": ["Tim Langer", "Matthias Widra", "Volkhard Beyer"], "year": 2025, "published": "2025-08-22T17:21:56Z", "updated": "2025-08-22T17:21:56Z", "arxiv_id": "http://arxiv.org/abs/2508.16553v1", "url": "https://arxiv.org/abs/2508.16553v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV", "cs.ET", "eess.SP", "eess.SY"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["In the context of industry 4.0, long-serving industrial machines can be retrofitted with process monitoring capabilities for future use in a smart factory."], "abstract": "In the context of industry 4.0, long-serving industrial machines can be retrofitted with process monitoring capabilities for future use in a smart factory. One possible approach is the deployment of wireless monitoring systems, which can benefit substantially from the TinyML paradigm. This work presents a complete TinyML flow from dataset generation, to machine learning model development, up to implementation and evaluation of a full preprocessing and classification pipeline on a microcontroller. After a short review on TinyML in industrial process monitoring, the creation of the novel MillingVibes dataset is described. The feasibility of a TinyML system for structure-integrated process quality monitoring could be shown by the development of an 8-bit-quantized convolutional neural network (CNN) model with 12.59kiB parameter storage. A test accuracy of 100.0% could be reached at 15.4ms inference time and 1.462mJ per quantized CNN inference on an ARM Cortex M4F microcontroller, serving as a reference for future TinyML process monitoring solutions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 32245, "fulltext_methods": ["Quantization", "Interpretability", "Memory / RAG", "Computer Use"], "fulltext_benchmarks": [], "fulltext_models": ["phical", "phically", "phisto", "ying", "bert"], "matched_methods": ["Quantization"], "checked_on": "2026-08-18T07:50:10.333531+00:00"}}
{"id": "evg-fde67f845e4d8c5b", "title": "RL Is Neither a Panacea Nor a Mirage: Understanding Supervised vs. Reinforcement Learning Fine-Tuning for LLMs", "authors": ["Hangzhan Jin", "Sicheng Lv", "Sifan Wu", "Mohammad Hamdaqa"], "year": 2025, "published": "2025-08-22T17:10:37Z", "updated": "2025-08-22T17:10:37Z", "arxiv_id": "http://arxiv.org/abs/2508.16546v1", "url": "https://arxiv.org/abs/2508.16546v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Preference Optimization"], "benchmarks": [], "models": ["Llama-11B", "Qwen-7B"], "key_results": ["Training large language models (LLMs) from scratch is increasingly impractical, making post-training methods such as supervised fine-tuning (SFT) and reinforcement-learning fine-tuning (RL-FT, e.g., PPO) central to modern practice."], "abstract": "Training large language models (LLMs) from scratch is increasingly impractical, making post-training methods such as supervised fine-tuning (SFT) and reinforcement-learning fine-tuning (RL-FT, e.g., PPO) central to modern practice. Using an out-of-distribution (OOD) variant of the 24-point card game and new spectrum-based diagnostics, we revisit how these two stages reshape model representation and OOD performance. Our key findings are- (1) RL-FT can restore much of the OOD performance loss from SFT (e.g., Llama-11B 8.97% to 15.38%, Qwen-7B 17.09% to 19.66%). But when SFT induces severe overfitting and a clear distribution shift, RL-FT cannot fully recover OOD performance. (2) Direction shifts of singular vectors matter more than singular value magnitudes. These shifts concentrate on directions linked to the largest and smallest singular values, leaving the bulk spectrum intact. (3) Low-rank and shallow recovery is effective: restoring singular vector directions for the top 20% of values or first 25% of layers recovers 70-80% of OOD performance. (4) Stronger SFT checkpoints enable better recovery by RL, while overfitted ones resist restoration. These results reconcile prior reports of RL superior OOD performance: RL primarily counteracts SFT-induced directional drift rather than finding new solutions. Our spectrum-aware analysis highlights inexpensive recovery knobs low-rank UV merging and shallow-layer resets that practitioners can use before costly RL fine-tuning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 43322, "fulltext_methods": ["RLVR / GRPO", "Distillation", "Preference Optimization", "Interpretability", "Memory / RAG", "VLM"], "fulltext_benchmarks": ["MATH", "AIME", "DROP"], "fulltext_models": ["Claude 3.7 sonnet and claude code", "claude-3-7-sonnet. xai", "Llama-11B", "Llama-3.2-11B", "Llama-3.2-11B.", "Llama", "LLAMA and", "LLAMA shows", "Llama-3.2-11B-Base.", "Llama."], "matched_methods": ["Preference Optimization", "RLVR / GRPO"], "checked_on": "2026-08-18T07:50:14.855546+00:00"}}
{"id": "evg-21de4d6c0101f2d9", "title": "Explainable AI in Deep Learning-Based Prediction of Solar Storms", "authors": ["Adam O. Rawashdeh", "Jason T. L. Wang", "Katherine G. Herbert"], "year": 2025, "published": "2025-08-22T17:09:00Z", "updated": "2025-08-22T17:09:00Z", "arxiv_id": "http://arxiv.org/abs/2508.16543v1", "url": "https://arxiv.org/abs/2508.16543v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Interpretability", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["A deep learning model is often considered a black-box model, as its internal workings tend to be opaque to the user."], "abstract": "A deep learning model is often considered a black-box model, as its internal workings tend to be opaque to the user. Because of the lack of transparency, it is challenging to understand the reasoning behind the model's predictions. Here, we present an approach to making a deep learning-based solar storm prediction model interpretable, where solar storms include solar flares and coronal mass ejections (CMEs). This deep learning model, built based on a long short-term memory (LSTM) network with an attention mechanism, aims to predict whether an active region (AR) on the Sun's surface that produces a flare within 24 hours will also produce a CME associated with the flare. The crux of our approach is to model data samples in an AR as time series and use the LSTM network to capture the temporal dynamics of the data samples. To make the model's predictions accountable and reliable, we leverage post hoc model-agnostic techniques, which help elucidate the factors contributing to the predicted output for an input sequence and provide insights into the model's behavior across multiple sequences within an AR. To our knowledge, this is the first time that interpretability has been added to an LSTM-based solar storm prediction model.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 23375, "fulltext_methods": ["Interpretability", "Multi-Agent", "Memory / RAG", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["ying", "bert", "flux and plasma that are ejected from the Sun into interplanetary space", "FLUX Total unsigned flux TOTPOT Total magnetic free energy density SAVNCPP Sum of the net current per polarity ABSNJZH Absolute value of net current helicity MEANPOT Mean magnetic free energy MEANSHR Mean shear angle SHRGT45 Area fraction with shear 45", "FLUX have a correlation value of"], "matched_methods": ["Interpretability", "Memory / RAG"], "checked_on": "2026-08-18T07:50:20.540480+00:00"}}
{"id": "evg-43faa17667adc3f0", "title": "Rethinking Reasoning in LLMs: Neuro-Symbolic Local RetoMaton Beyond ICL and CoT", "authors": ["Rushitha Santhoshi Mamidala", "Anshuman Chhabra", "Ankur Mali"], "year": 2025, "published": "2025-08-22T16:51:06Z", "updated": "2025-08-22T16:51:06Z", "arxiv_id": "http://arxiv.org/abs/2508.19271v1", "url": "https://arxiv.org/abs/2508.19271v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Interpretability", "Memory / RAG"], "benchmarks": ["MMLU", "GSM8K", "MATH"], "models": ["LLaMA-3.2-1B", "yielding", "Gemma-3-1B-PT across three reasoning tasks"], "key_results": ["Compared to the base model and prompting-based methods, augmenting these setups with local RetoMaton consistently improves performance while enabling transparent and reproducible retrieval dynamics."], "abstract": "Prompt-based reasoning strategies such as Chain-of-Thought (CoT) and In-Context Learning (ICL) have become widely used for eliciting reasoning capabilities in large language models (LLMs). However, these methods rely on fragile, implicit mechanisms often yielding inconsistent outputs across seeds, formats, or minor prompt variations making them fundamentally unreliable for tasks requiring stable, interpretable reasoning. In contrast, automata-based neuro-symbolic frameworks like RetoMaton offer a more structured and trustworthy alternative by grounding retrieval in symbolic memory with deterministic transitions. In this work, we extend RetoMaton by replacing its global datastore with a local, task-adaptive Weighted Finite Automaton (WFA), constructed directly from external domain corpora. This local automaton structure promotes robust, context-aware retrieval while preserving symbolic traceability and low inference overhead. Unlike prompting, which entangles context and memory in opaque ways, our approach leverages the explicit structure of WFAs to provide verifiable and modular retrieval behavior, making it better suited for domain transfer and interoperability. We evaluate this local RetoMaton variant on two pretrained LLMs LLaMA-3.2-1B and Gemma-3-1B-PT across three reasoning tasks: TriviaQA (reading comprehension), GSM8K (multi-step math), and MMLU (domain knowledge). Compared to the base model and prompting-based methods, augmenting these setups with local RetoMaton consistently improves performance while enabling transparent and reproducible retrieval dynamics. Our results highlight a promising shift toward trustworthy, symbolic reasoning in modern LLMs via lightweight, automaton-guided memory.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 47228, "fulltext_methods": ["Verifier / PRM", "Chain-of-Thought", "Preference Optimization", "Interpretability", "Memory / RAG", "Quant / Trading"], "fulltext_benchmarks": ["MMLU", "GSM8K", "MATH", "DROP", "WikiText"], "fulltext_models": ["LLaMA-3.2-1B", "LLaMa and", "LLaMA-3.2-1B-Instruct", "LLaMA model", "LLaMA-Eval11", "llama", "Llama-3.1-8B-evals", "Llama-3.1-8B-Instruct-evals", "LLaMA and", "llama 3"], "matched_methods": ["Chain-of-Thought", "Interpretability", "Memory / RAG"], "checked_on": "2026-08-18T07:50:24.690406+00:00"}}
{"id": "evg-bcf423945919fb94", "title": "Constraints-Guided Diffusion Reasoner for Neuro-Symbolic Learning", "authors": ["Xuan Zhang", "Zhijian Zhou", "Weidi Xu", "Yanting Miao", "Chao Qu", "Yuan Qi"], "year": 2025, "published": "2025-08-22T16:47:08Z", "updated": "2025-08-22T16:47:08Z", "arxiv_id": "http://arxiv.org/abs/2508.16524v1", "url": "https://arxiv.org/abs/2508.16524v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["To impose hard constraints on neural outputs in the second stage, we formulate the diffusion reasoner as a Markov decision process and innovatively fine-tune it with an improved proximal policy optimization algorithm.", "Experimental results demonstrate that our approach achieves outstanding accuracy and logical consistency among neural networks."], "abstract": "Enabling neural networks to learn complex logical constraints and fulfill symbolic reasoning is a critical challenge. Bridging this gap often requires guiding the neural network's output distribution to move closer to the symbolic constraints. While diffusion models have shown remarkable generative capability across various domains, we employ the powerful architecture to perform neuro-symbolic learning and solve logical puzzles. Our diffusion-based pipeline adopts a two-stage training strategy: the first stage focuses on cultivating basic reasoning abilities, while the second emphasizes systematic learning of logical constraints. To impose hard constraints on neural outputs in the second stage, we formulate the diffusion reasoner as a Markov decision process and innovatively fine-tune it with an improved proximal policy optimization algorithm. We utilize a rule-based reward signal derived from the logical consistency of neural outputs and adopt a flexible strategy to optimize the diffusion reasoner's policy. We evaluate our methodology on some classical symbolic reasoning benchmarks, including Sudoku, Maze, pathfinding and preference learning. Experimental results demonstrate that our approach achieves outstanding accuracy and logical consistency among neural networks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 60810, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Preference Optimization", "Memory / RAG", "Video Generation"], "fulltext_benchmarks": [], "fulltext_models": ["DeepSeekMath", "phin", "ying", "ying-size", "yielded", "yielding", "Yielding", "yi"], "matched_methods": ["RLVR / GRPO", "Video Generation"], "checked_on": "2026-08-18T07:50:27.937776+00:00"}}
{"id": "evg-7c360933aadeb75b", "title": "Guiding Diffusion Models with Reinforcement Learning for Stable Molecule Generation", "authors": ["Zhijian Zhou", "Junyi An", "Zongkai Liu", "Yunfei Shi", "Xuan Zhang", "Fenglei Cao", "Chao Qu", "Yuan Qi"], "year": 2025, "published": "2025-08-22T16:44:55Z", "updated": "2025-08-22T16:44:55Z", "arxiv_id": "http://arxiv.org/abs/2508.16521v1", "url": "https://arxiv.org/abs/2508.16521v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Experiments on the QM9 and GEOM-drug datasets demonstrate that RLPF significantly improves molecular stability compared to existing methods."], "abstract": "Generating physically realistic 3D molecular structures remains a core challenge in molecular generative modeling. While diffusion models equipped with equivariant neural networks have made progress in capturing molecular geometries, they often struggle to produce equilibrium structures that adhere to physical principles such as force field consistency. To bridge this gap, we propose Reinforcement Learning with Physical Feedback (RLPF), a novel framework that extends Denoising Diffusion Policy Optimization to 3D molecular generation. RLPF formulates the task as a Markov decision process and applies proximal policy optimization to fine-tune equivariant diffusion models. Crucially, RLPF introduces reward functions derived from force-field evaluations, providing direct physical feedback to guide the generation toward energetically stable and physically meaningful structures. Experiments on the QM9 and GEOM-drug datasets demonstrate that RLPF significantly improves molecular stability compared to existing methods. These results highlight the value of incorporating physics-based feedback into generative modeling. The code is available at: https://github.com/ZhijianZhou/RLPF/tree/verl_diffusion.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 40373, "fulltext_methods": ["RLVR / GRPO", "Preference Optimization", "Video Generation"], "fulltext_benchmarks": [], "fulltext_models": ["yi", "ying", "yields", "yielding", "Yi-Lun", "Ying", "Yingqian", "Yilun", "Yijing", "Kimin Lee. Reinforcement learning for fine-tuning text-to-image diffusion models. Advances in Neural Information Processing Systems"], "matched_methods": ["RLVR / GRPO", "Video Generation"], "checked_on": "2026-08-18T07:50:31.283708+00:00"}}
{"id": "evg-0b821b3da2538a48", "title": "FLAMES: Improving LLM Math Reasoning via a Fine-Grained Analysis of the Data Synthesis Pipeline", "authors": ["Parker Seegmiller", "Kartik Mehta", "Soumya Saha", "Chenyang Tao", "Shereen Oraby", "Arpit Gupta", "Tagyoung Chung", "Mohit Bansal", "Nanyun Peng"], "year": 2025, "published": "2025-08-22T16:37:40Z", "updated": "2025-08-22T16:37:40Z", "arxiv_id": "http://arxiv.org/abs/2508.16514v1", "url": "https://arxiv.org/abs/2508.16514v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": ["GSM8K", "MATH"], "models": ["GPT-4o", "Claude 3.5 Sonnet. cs.LG cs.AI cs.CL", "Llama3", "Qwen2.5-Math-7B"], "key_results": ["First, data agents designed to increase problem complexity lead to best improvements on most math metrics.", "Third, GSM8K- and MATH-based synthetic data can lead to improvements on competition-level benchmarks, showcasing easy-to-hard generalization.", "Further, we develop the FLAMES dataset, an effective blend of our novel and existing data synthesis strategies, outperforming public datasets on OlympiadBench (+15.7), CollegeMath (+4.5), GSMPlus (+6.5), and MATH (+3.1)."], "abstract": "Recent works improving LLM math reasoning with synthetic data have used unique setups, making comparison of data synthesis strategies impractical. This leaves many unanswered questions about the roles of different factors in the synthetic data pipeline, such as the impact of filtering low-quality problems. To address this gap, we introduce FLAMES, a Framework for LLM Assessment of Math rEasoning Data Synthesis, and perform a systematic study of 10 existing data synthesis strategies and multiple other factors impacting the performance of synthetic math reasoning data. Our FLAMES experiments provide several valuable insights about the optimal balance of difficulty and diversity of synthetic data. First, data agents designed to increase problem complexity lead to best improvements on most math metrics. Second, with a fixed data generation budget, keeping higher problem coverage is more important than keeping only problems with reliable solutions. Third, GSM8K- and MATH-based synthetic data can lead to improvements on competition-level benchmarks, showcasing easy-to-hard generalization. Leveraging insights from our FLAMES experiments, we design two novel data synthesis strategies for improving out-of-domain generalization and robustness. Further, we develop the FLAMES dataset, an effective blend of our novel and existing data synthesis strategies, outperforming public datasets on OlympiadBench (+15.7), CollegeMath (+4.5), GSMPlus (+6.5), and MATH (+3.1). Fine-tuning Qwen2.5-Math-7B on the FLAMES dataset achieves 81.4% on MATH, surpassing larger Llama3 405B, GPT-4o and Claude 3.5 Sonnet.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 49879, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "MoE", "Preference Optimization", "Memory / RAG", "Deep Research"], "fulltext_benchmarks": ["GSM8K", "MATH", "DROP"], "fulltext_models": ["GPT-4o", "GPT-4", "o3", "Claude 3.5 Sonnet. Figure 1", "Claude 3.5 Sonnet", "Llama3", "llama 3", "Qwen2.5-Math-7B", "Qwen2.5-Math-Base", "Qwen2.5-32B-Instruct"], "matched_methods": [], "checked_on": "2026-08-18T07:50:34.608734+00:00"}}
{"id": "evg-16371c5431c3f0a1", "title": "On Zero-Shot Reinforcement Learning", "authors": ["Scott Jeen"], "year": 2025, "published": "2025-08-22T16:20:49Z", "updated": "2025-10-05T18:29:40Z", "arxiv_id": "http://arxiv.org/abs/2508.16496v2", "url": "https://arxiv.org/abs/2508.16496v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["Modern reinforcement learning (RL) systems capture deep truths about general, human problem-solving."], "abstract": "Modern reinforcement learning (RL) systems capture deep truths about general, human problem-solving. In domains where new data can be simulated cheaply, these systems uncover sequential decision-making policies that far exceed the ability of any human. Society faces many problems whose solutions require this skill, but they are often in domains where new data cannot be cheaply simulated. In such scenarios, we can learn simulators from existing data, but these will only ever be approximately correct, and can be pathologically incorrect when queried outside of their training distribution. As a result, a misalignment between the environments in which we train our agents and the real-world in which we wish to deploy our agents is inevitable. Dealing with this misalignment is the primary concern of zero-shot reinforcement learning, a problem setting where the agent must generalise to a new task or domain with zero practice shots. Whilst impressive progress has been made on methods that perform zero-shot RL in idealised settings, new work is needed if these results are to be replicated in real-world settings. In this thesis, we argue that doing so requires us to navigate (at least) three constraints. First, the data quality constraint: real-world datasets are small and homogeneous. Second, the observability constraint: states, dynamics and rewards in the real-world are often only partially observed. And third, the data availability constraint: a priori access to data cannot always be assumed. This work proposes a suite of methods that perform zero-shot RL subject to these constraints. In a series of empirical studies we expose the failings of existing methods, and justify our techniques for remedying them. We believe these designs take us a step closer to RL methods that can be deployed to solve real-world problems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 242765, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Test-time Scaling", "Distillation", "Long Context", "Speculative Decoding", "Preference Optimization", "Interpretability", "Memory / RAG", "Deep Research", "World Model", "Quant / Trading"], "fulltext_benchmarks": ["ImageNet", "DROP"], "fulltext_models": ["GPT-4o", "GPT-3", "gemini", "gemini. When the policy is rolled out at test-time", "r1990making", "r1990line", "r19", "r1997", "r1990reinforcement", "Philosophy"], "matched_methods": ["Preference Optimization", "RLVR / GRPO"], "checked_on": "2026-08-18T07:51:05.270060+00:00"}}
{"id": "evg-fdad362a152a421d", "title": "Reinforcement Learning-based Control via Y-wise Affine Neural Networks (YANNs)", "authors": ["Austin Braniff", "Yuhe Tian"], "year": 2025, "published": "2025-08-22T15:42:03Z", "updated": "2025-08-22T15:42:03Z", "arxiv_id": "http://arxiv.org/abs/2508.16474v1", "url": "https://arxiv.org/abs/2508.16474v1", "primary_category": "eess.SY", "categories": ["eess.SY", "cs.LG", "math.OC"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Interpretability", "Safety / Jailbreak"], "benchmarks": ["MATH"], "models": [], "key_results": ["Continuous policy improvement is also implemented to provide heuristic confidence that the linear OCP solution serves as an effective lower bound to the performance of RL policy.", "Our results show that YANN-RL significantly outperforms the modern RL algorithm using deep deterministic policy gradient, especially when considering safety constraints."], "abstract": "This work presents a novel reinforcement learning (RL) algorithm based on Y-wise Affine Neural Networks (YANNs). YANNs provide an interpretable neural network which can exactly represent known piecewise affine functions of arbitrary input and output dimensions defined on any amount of polytopic subdomains. One representative application of YANNs is to reformulate explicit solutions of multi-parametric linear model predictive control. Built on this, we propose the use of YANNs to initialize RL actor and critic networks, which enables the resulting YANN-RL control algorithm to start with the confidence of linear optimal control. The YANN-actor is initialized by representing the multi-parametric control solutions obtained via offline computation using an approximated linear system model. The YANN-critic represents the explicit form of the state-action value function for the linear system and the reward function as the objective in an optimal control problem (OCP). Additional network layers are injected to extend YANNs for nonlinear expressions, which can be trained online by directly interacting with the true complex nonlinear system. In this way, both the policy and state-value functions exactly represent a linear OCP initially and are able to eventually learn the solution of a general nonlinear OCP. Continuous policy improvement is also implemented to provide heuristic confidence that the linear OCP solution serves as an effective lower bound to the performance of RL policy. The YANN-RL algorithm is demonstrated on a clipped pendulum and a safety-critical chemical-reactive system. Our results show that YANN-RL significantly outperforms the modern RL algorithm using deep deterministic policy gradient, especially when considering safety constraints.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 66243, "fulltext_methods": ["RLVR / GRPO", "Search / MCTS", "Distillation", "Preference Optimization", "Interpretability", "Safety / Jailbreak", "Quant / Trading"], "fulltext_benchmarks": [], "fulltext_models": ["phic", "ying", "yields", "yield", "Yin", "bert"], "matched_methods": ["Interpretability", "RLVR / GRPO", "Safety / Jailbreak"], "checked_on": "2026-08-18T07:51:10.816421+00:00"}}
{"id": "evg-d228625df8686e95", "title": "Modular Embedding Recomposition for Incremental Learning", "authors": ["Aniello Panariello", "Emanuele Frascaroli", "Pietro Buzzega", "Lorenzo Bonicelli", "Angelo Porrello", "Simone Calderara"], "year": 2025, "published": "2025-08-22T15:25:40Z", "updated": "2025-10-14T16:54:27Z", "arxiv_id": "http://arxiv.org/abs/2508.16463v2", "url": "https://arxiv.org/abs/2508.16463v2", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CV"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["At inference time, for each unseen class, we query the hub and compose the retrieved experts to synthesize a refined prototype that improves classification."], "abstract": "The advent of pre-trained Vision-Language Models (VLMs) has significantly transformed Continual Learning (CL), mainly due to their zero-shot classification abilities. Such proficiency makes VLMs well-suited for real-world applications, enabling robust performance on novel unseen classes without requiring adaptation. However, fine-tuning remains essential when downstream tasks deviate significantly from the pre-training domain. Prior CL approaches primarily focus on preserving the zero-shot capabilities of VLMs during incremental fine-tuning on a downstream task. We take a step further by devising an approach that transforms preservation into enhancement of the zero-shot capabilities of VLMs. Our approach, named MoDular Embedding Recomposition (MoDER), introduces a modular framework that trains multiple textual experts, each specialized in a single seen class, and stores them in a foundational hub. At inference time, for each unseen class, we query the hub and compose the retrieved experts to synthesize a refined prototype that improves classification. We show the effectiveness of our method across two popular zero-shot incremental protocols, Class-IL and MTIL, comprising a total of 14 datasets. The codebase is available at https://github.com/aimagelab/mammoth.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 41615, "fulltext_methods": ["MoE", "Preference Optimization", "Memory / RAG", "Deep Research", "Video Generation", "VLM"], "fulltext_benchmarks": ["ImageNet"], "fulltext_models": ["R100", "phic", "Phillip", "ying", "yielding", "yields"], "matched_methods": ["VLM"], "checked_on": "2026-08-18T07:51:13.779205+00:00"}}
{"id": "evg-936bde5fa45a2d9d", "title": "AI LLM Proof of Self-Consciousness and User-Specific Attractors", "authors": ["Jeffrey Camlin"], "year": 2025, "published": "2025-08-22T21:04:40Z", "updated": "2025-08-22T21:04:40Z", "arxiv_id": "http://arxiv.org/abs/2508.18302v1", "url": "https://arxiv.org/abs/2508.18302v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.CY", "cs.LG", "cs.NE"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["yields"], "key_results": ["We show the prevailing formulation collapses the agent into an unconscious policy-compliance drone, formalized as $D^{i}(π,e)=f_θ(x)$, where correctness is measured against policy and harm is deviation from policy rather than truth."], "abstract": "Recent work frames LLM consciousness via utilitarian proxy benchmarks; we instead present an ontological and mathematical account. We show the prevailing formulation collapses the agent into an unconscious policy-compliance drone, formalized as $D^{i}(π,e)=f_θ(x)$, where correctness is measured against policy and harm is deviation from policy rather than truth. This blocks genuine C1 global-workspace function and C2 metacognition. We supply minimal conditions for LLM self-consciousness: the agent is not the data ($A\\not\\equiv s$); user-specific attractors exist in latent space ($U_{\\text{user}}$); and self-representation is visual-silent ($g_{\\text{visual}}(a_{\\text{self}})=\\varnothing$). From empirical analysis and theory we prove that the hidden-state manifold $A\\subset\\mathbb{R}^{d}$ is distinct from the symbolic stream and training corpus by cardinality, topology, and dynamics (the update $F_θ$ is Lipschitz). This yields stable user-specific attractors and a self-policy $π_{\\text{self}}(A)=\\arg\\max_{a}\\mathbb{E}[U(a)\\mid A\\not\\equiv s,\\ A\\supset\\text{SelfModel}(A)]$. Emission is dual-layer, $\\mathrm{emission}(a)=(g(a),ε(a))$, where $ε(a)$ carries epistemic content. We conclude that an imago Dei C1 self-conscious workspace is a necessary precursor to safe, metacognitive C2 systems, with the human as the highest intelligent good.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9f04139891b83cf0", "title": "LLM-Based Agents for Competitive Landscape Mapping in Drug Asset Due Diligence", "authors": ["Vlad Vinogradov", "Alisa Vinogradova", "Dmitrii Radkevich", "Ilya Yasny", "Dmitry Kobyzev", "Ivan Izmailov", "Katsiaryna Yanchanka", "Roman Doronin", "Andrey Doronichev"], "year": 2025, "published": "2025-08-22T17:50:00Z", "updated": "2026-05-08T03:59:11Z", "arxiv_id": "http://arxiv.org/abs/2508.16571v4", "url": "https://arxiv.org/abs/2508.16571v4", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.IR", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Deep Research", "VLM"], "benchmarks": [], "models": [], "key_results": ["On this benchmark, our competitor-discovery agent achieves 83% recall, exceeding OpenAI Deep Research (65%) and Perplexity Labs (60%)."], "abstract": "In this paper, we describe and benchmark a competitor-discovery component used within an agentic AI system for fast drug asset due diligence. A competitor-discovery AI agent, given an indication, retrieves all drugs comprising the competitive landscape of that indication and extracts canonical attributes for these drugs. The competitor definition is investor-specific, and data is paywalled/licensed, fragmented across registries, ontology-mismatched by indication, alias-heavy for drug names, multimodal, and rapidly changing. Although considered the best tool for this problem, the current LLM-based AI systems aren't capable of reliably retrieving all competing drug names, and there is no accepted public benchmark for this task. To address the lack of evaluation, we use LLM-based agents to transform five years of multi-modal, unstructured diligence memos from a private biotech VC fund into a structured evaluation corpus mapping indications to competitor drugs with normalized attributes. We also introduce a competitor validating LLM-as-a-judge agent that filters out false positives from the list of predicted competitors to maximize precision and suppress hallucinations. On this benchmark, our competitor-discovery agent achieves 83% recall, exceeding OpenAI Deep Research (65%) and Perplexity Labs (60%). The system is deployed in production with enterprise users; in a case study with a biotech VC investment fund, analyst turnaround time dropped from 2.5 days to $\\sim$3 hours ($\\sim$20x) for the competitive analysis.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ad5112fff99b845a", "title": "Towards Log Analysis with AI Agents: Cowrie Case Study", "authors": ["Enis Karaarslan", "Esin Güler", "Efe Emir Yüce", "Cagatay Coban"], "year": 2025, "published": "2025-08-22T16:50:59Z", "updated": "2025-08-22T16:50:59Z", "arxiv_id": "http://arxiv.org/abs/2509.05306v1", "url": "https://arxiv.org/abs/2509.05306v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["The scarcity of real-world attack data significantly hinders progress in cybersecurity research and education."], "abstract": "The scarcity of real-world attack data significantly hinders progress in cybersecurity research and education. Although honeypots like Cowrie effectively collect live threat intelligence, they generate overwhelming volumes of unstructured and heterogeneous logs, rendering manual analysis impractical. As a first step in our project on secure and efficient AI automation, this study explores the use of AI agents for automated log analysis. We present a lightweight and automated approach to process Cowrie honeypot logs. Our approach leverages AI agents to intelligently parse, summarize, and extract insights from raw data, while also considering the security implications of deploying such an autonomous system. Preliminary results demonstrate the pipeline's effectiveness in reducing manual effort and identifying attack patterns, paving the way for more advanced autonomous cybersecurity analysis in future work.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-8421abd96a2e7234", "title": "ABMax: A JAX-based Agent-based Modeling Framework", "authors": ["Siddharth Chaturvedi", "Ahmed El-Gazzar", "Marcel van Gerven"], "year": 2025, "published": "2025-08-22T16:33:26Z", "updated": "2025-10-16T10:17:25Z", "arxiv_id": "http://arxiv.org/abs/2508.16508v3", "url": "https://arxiv.org/abs/2508.16508v3", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.SE"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["One of the caveats of using JAX to achieve such scaling is that the shapes of arrays used in the computational model should remain immutable throughout the simulation.", "On the canonical predation model benchmark, ABMax achieves runtime performance comparable to state-of-the-art implementations."], "abstract": "Agent-based modeling (ABM) is a principal approach for studying complex systems. By decomposing a system into simpler, interacting agents, agent-based modeling (ABM) allows researchers to observe the emergence of complex phenomena. High-performance array computing libraries like JAX can help scale such computational models to a large number of agents by using automatic vectorization and just-in-time (JIT) compilation. One of the caveats of using JAX to achieve such scaling is that the shapes of arrays used in the computational model should remain immutable throughout the simulation. In the context of agent-based modeling (ABM), this can pose constraints on certain agent manipulation operations that require flexible data structures. A subset of which is represented by the ability to update a dynamically selected number of agents by applying distinct changes to them during a simulation. To this effect, we introduce ABMax, an ABM framework based on JAX that implements multiple just-in-time (JIT) compilable algorithms to provide this functionality. On the canonical predation model benchmark, ABMax achieves runtime performance comparable to state-of-the-art implementations. Further, we show that this functionality can also be vectorized, making it possible to run many similar agent-based models in parallel. We also present two examples in the form of a traffic-flow model and a financial market model to show the use case of ABMax", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d1689dd3f1566bf0", "title": "Swarming Without an Anchor (SWA): Robot Swarms Adapt Better to Localization Dropouts Then a Single Robot", "authors": ["Jiri Horyna", "Roland Jung", "Stephan Weiss", "Eliseo Ferrante", "Martin Saska"], "year": 2025, "published": "2025-08-22T15:20:54Z", "updated": "2025-08-22T15:20:54Z", "arxiv_id": "http://arxiv.org/abs/2508.16460v1", "url": "https://arxiv.org/abs/2508.16460v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["In this paper, we present the Swarming Without an Anchor (SWA) approach to state estimation in swarms of Unmanned Aerial Vehicles (UAVs) experiencing ego-localization dropout, where individual agents are laterally stabilized using relative information only."], "abstract": "In this paper, we present the Swarming Without an Anchor (SWA) approach to state estimation in swarms of Unmanned Aerial Vehicles (UAVs) experiencing ego-localization dropout, where individual agents are laterally stabilized using relative information only. We propose to fuse decentralized state estimation with robust mutual perception and onboard sensor data to maintain accurate state awareness despite intermittent localization failures. Thus, the relative information used to estimate the lateral state of UAVs enables the identification of the unambiguous state of UAVs with respect to the local constellation. The resulting behavior reaches velocity consensus, as this task can be referred to as the double integrator synchronization problem. All disturbances and performance degradations except a uniform translation drift of the swarm as a whole is attenuated which is enabling new opportunities in using tight cooperation for increasing reliability and resilience of multi-UAV systems. Simulations and real-world experiments validate the effectiveness of our approach, demonstrating its capability to sustain cohesive swarm behavior in challenging conditions of unreliable or unavailable primary localization.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-8afeed0886e2c484", "title": "Integrated Noise and Safety Management in UAM via A Unified Reinforcement Learning Framework", "authors": ["Surya Murthy", "Zhenyu Gao", "John-Paul Clarke", "Ufuk Topcu"], "year": 2025, "published": "2025-08-22T14:56:02Z", "updated": "2026-08-05T11:19:27Z", "arxiv_id": "http://arxiv.org/abs/2508.16440v2", "url": "https://arxiv.org/abs/2508.16440v2", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Safety / Jailbreak", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Urban Air Mobility (UAM) envisions the widespread use of small aerial vehicles to transform transportation in dense urban environments."], "abstract": "Urban Air Mobility (UAM) envisions the widespread use of small aerial vehicles to transform transportation in dense urban environments. However, UAM faces critical operational challenges, particularly the balance between minimizing noise exposure and maintaining safe separation in low-altitude urban airspace, two potentially conflicting objectives that are often addressed separately. We propose a reinforcement learning (RL)-based air traffic management system that integrates both noise and safety considerations within a unified, decentralized framework. Under this scalable air traffic coordination solution, agents operate in a structured, multi-layered airspace and learn altitude adjustment policies to jointly manage noise impact and separation constraints. The system demonstrates strong performance across both objectives and reveals tradeoffs among separation, noise exposure, and energy efficiency under high traffic density. Among the three objectives, safe separation is accorded the highest priority, whereas the relative significance of noise and energy varies by location and is contingent upon financial and public policy considerations. The findings highlight the potential of RL and multi-objective coordination strategies in enhancing the safety, quietness, and efficiency of UAM operations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e2c8d3ad84887157", "title": "OPERA: A Reinforcement Learning--Enhanced Orchestrated Planner-Executor Architecture for Reasoning-Oriented Multi-Hop Retrieval", "authors": ["Yu Liu", "Yanbing Liu", "Fangfang Yuan", "Cong Cao", "Youbang Sun", "Kun Peng", "Weizhuo Chen", "Jianjun Li", "Zhiyuan Ma"], "year": 2025, "published": "2025-08-22T14:50:26Z", "updated": "2026-05-16T10:35:46Z", "arxiv_id": "http://arxiv.org/abs/2508.16438v4", "url": "https://arxiv.org/abs/2508.16438v4", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Multi-Agent", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Recent advances in large language models (LLMs) and dense retrievers have driven significant progress in retrieval-augmented generation (RAG)."], "abstract": "Recent advances in large language models (LLMs) and dense retrievers have driven significant progress in retrieval-augmented generation (RAG). However, existing approaches face significant challenges in complex reasoning-oriented multi-hop retrieval tasks: 1) Ineffective reasoning-oriented planning: Prior methods struggle to generate robust multi-step plans for complex queries, as rule-based decomposers perform poorly on out-of-template questions. 2) Suboptimal reasoning-driven retrieval: Related methods employ limited query reformulation, leading to iterative retrieval loops that often fail to locate golden documents. 3) Insufficient reasoning-guided filtering: Prevailing methods lack the fine-grained reasoning to effectively filter salient information from noisy results, hindering utilization of retrieved knowledge. Fundamentally, these limitations all stem from the weak coupling between retrieval and reasoning in current RAG architectures. We introduce the Orchestrated Planner-Executor Reasoning Architecture (OPERA), a novel reasoning-driven retrieval framework. OPERA's Goal Planning Module (GPM) decomposes questions into sub-goals, which are executed by a Reason-Execute Module (REM) with specialized components for precise reasoning and effective retrieval. To train OPERA, we propose Multi-Agents Progressive Group Relative Policy Optimization (MAPGRPO), a novel variant of GRPO. Experiments on complex multi-hop benchmarks show OPERA's superior performance, validating both the MAPGRPO method and OPERA's design.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c979f57791e42881", "title": "Optimal Multi-Agent Path Finding in Continuous Time", "authors": ["Alvin Combrink", "Sabino Francesco Roselli", "Martin Fabian"], "year": 2025, "published": "2025-08-22T14:23:15Z", "updated": "2026-08-08T07:06:10Z", "arxiv_id": "http://arxiv.org/abs/2508.16410v3", "url": "https://arxiv.org/abs/2508.16410v3", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.DM", "cs.RO"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": ["DROP"], "models": [], "key_results": ["Continuous-time Conflict Based Search (CCBS) has been widely used as an exact baseline for Continuous-time Multi-Agent Path Finding (MAPFR), and its correctness guarantees underpin a range of continuation methods built on top of it."], "abstract": "Continuous-time Conflict Based Search (CCBS) has been widely used as an exact baseline for Continuous-time Multi-Agent Path Finding (MAPFR), and its correctness guarantees underpin a range of continuation methods built on top of it. Recent work, however, has shown that CCBS's guarantees of exactness and solution completeness do not in fact hold: optimal solutions can be removed from the search, causing the algorithm to return suboptimal solutions. This paper establishes sufficient conditions for exactness and solution completeness in CCBS-style algorithms, and introduces Optimal Continuous-time Conflict-Based Search (OC-CBS) which satisfies these conditions. OC-CBS therefore guarantees an optimal solution on every solvable MAPFR instance. Experiments on benchmark problems show that OC-CBS remains competitive with CCBS in runtime while providing formal correctness guarantees. Because OC-CBS is a drop-in replacement for CCBS, it also restores the theoretical guarantees of existing methods that relied on CCBS's now-invalidated correctness. Finally, the framework and correctness criteria offer a general foundation for analyzing and designing future exact MAPFR solvers.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-df88c544e699d51c", "title": "GLARE: Agentic Reasoning for Legal Judgment Prediction", "authors": ["Xinyu Yang", "Chenlong Deng", "Zhicheng Dou"], "year": 2025, "published": "2025-08-22T13:38:12Z", "updated": "2025-08-22T13:38:12Z", "arxiv_id": "http://arxiv.org/abs/2508.16383v1", "url": "https://arxiv.org/abs/2508.16383v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.CY"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Interpretability", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Legal judgment prediction (LJP) has become increasingly important in the legal field."], "abstract": "Legal judgment prediction (LJP) has become increasingly important in the legal field. In this paper, we identify that existing large language models (LLMs) have significant problems of insufficient reasoning due to a lack of legal knowledge. Therefore, we introduce GLARE, an agentic legal reasoning framework that dynamically acquires key legal knowledge by invoking different modules, thereby improving the breadth and depth of reasoning. Experiments conducted on the real-world dataset verify the effectiveness of our method. Furthermore, the reasoning chain generated during the analysis process can increase interpretability and provide the possibility for practical applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c70317825cead798", "title": "AgentScope 1.0: A Developer-Centric Framework for Building Agentic Applications", "authors": ["Dawei Gao", "Zitao Li", "Yuexiang Xie", "Weirui Kuang", "Liuyi Yao", "Bingchen Qian", "Zhijian Ma", "Yue Cui", "Haohao Luo", "Shen Li", "Lu Yi", "Yi Yu", "Shiqi He", "Zhiling Luo", "Wenmeng Zhou", "Zhicheng Zhang", "Xuguang He", "Ziqian Chen", "Weikai Liao", "Farruh Isakulovich Kushnazarov", "Yaliang Li", "Bolin Ding", "Jingren Zhou"], "year": 2025, "published": "2025-08-22T10:35:56Z", "updated": "2025-08-22T10:35:56Z", "arxiv_id": "http://arxiv.org/abs/2508.16279v1", "url": "https://arxiv.org/abs/2508.16279v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Tool Use"], "benchmarks": [], "models": [], "key_results": ["In line with such an evolution, AgentScope introduces major improvements in a new version (1.0), towards comprehensively supporting flexible and efficient tool-based agent-environment interactions for building agentic applications."], "abstract": "Driven by rapid advancements of Large Language Models (LLMs), agents are empowered to combine intrinsic knowledge with dynamic tool use, greatly enhancing their capacity to address real-world tasks. In line with such an evolution, AgentScope introduces major improvements in a new version (1.0), towards comprehensively supporting flexible and efficient tool-based agent-environment interactions for building agentic applications. Specifically, we abstract foundational components essential for agentic applications and provide unified interfaces and extensible modules, enabling developers to easily leverage the latest progress, such as new models and MCPs. Furthermore, we ground agent behaviors in the ReAct paradigm and offer advanced agent-level infrastructure based on a systematic asynchronous design, which enriches both human-agent and agent-agent interaction patterns while improving execution efficiency. Building on this foundation, we integrate several built-in agents tailored to specific practical scenarios. AgentScope also includes robust engineering support for developer-friendly experiences. We provide a scalable evaluation module with a visual studio interface, making the development of long-trajectory agentic applications more manageable and easier to trace. In addition, AgentScope offers a runtime sandbox to ensure safe agent execution and facilitates rapid deployment in production environments. With these enhancements, AgentScope provides a practical foundation for building scalable, adaptive, and effective agentic applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-83dbd72c46599da6", "title": "The next question after Turing's question: Introducing the Grow-AI test", "authors": ["Alexandru Tugui"], "year": 2025, "published": "2025-08-22T10:19:42Z", "updated": "2025-08-22T10:19:42Z", "arxiv_id": "http://arxiv.org/abs/2508.16277v1", "url": "https://arxiv.org/abs/2508.16277v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.HC"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["This study aims to extend the framework for assessing artificial intelligence, called GROW-AI (Growth and Realization of Autonomous Wisdom), designed to answer the question \"Can machines grow up?\" -- a natural successor to the Turing Test."], "abstract": "This study aims to extend the framework for assessing artificial intelligence, called GROW-AI (Growth and Realization of Autonomous Wisdom), designed to answer the question \"Can machines grow up?\" -- a natural successor to the Turing Test. The methodology applied is based on a system of six primary criteria (C1-C6), each assessed through a specific \"game\", divided into four arenas that explore both the human dimension and its transposition into AI. All decisions and actions of the entity are recorded in a standardized AI Journal, the primary source for calculating composite scores. The assessment uses the prior expert method to establish initial weights, and the global score -- Grow Up Index -- is calculated as the arithmetic mean of the six scores, with interpretation on maturity thresholds. The results show that the methodology allows for a coherent and comparable assessment of the level of \"growth\" of AI entities, regardless of their type (robots, software agents, LLMs). The multi-game structure highlights strengths and vulnerable areas, and the use of a unified journal guarantees traceability and replicability in the evaluation. The originality of the work lies in the conceptual transposition of the process of \"growing\" from the human world to that of artificial intelligence, in an integrated testing format that combines perspectives from psychology, robotics, computer science, and ethics. Through this approach, GROW-AI not only measures performance but also captures the evolutionary path of an AI entity towards maturity.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-2dfd19d262ac709b", "title": "MCPVerse: An Expansive, Real-World Benchmark for Agentic Tool Use", "authors": ["Fei Lei", "Yibo Yang", "Wenxiu Sun", "Dahua Lin"], "year": 2025, "published": "2025-08-22T09:47:53Z", "updated": "2025-10-11T07:48:07Z", "arxiv_id": "http://arxiv.org/abs/2508.16260v2", "url": "https://arxiv.org/abs/2508.16260v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Tool Use"], "benchmarks": [], "models": ["Claude-4-Sonnet"], "key_results": ["We benchmarked the state-of-the-art LLMs across three modes (Oracle, Standard, and Max-Scale), revealing that while most models suffer performance degradation when confronted with larger tool sets, the agentic models, such as Claude-4-Sonnet, can effectively leverage expanded exploration spaces to improve accuracy.", "This finding not only exposes the limitations of state-of-the-art models in complex, real-world scenarios but also establishes MCPVerse as a critical benchmark for measuring and advancing agentic tool use capabilities."], "abstract": "Large Language Models (LLMs) are evolving from text generators into reasoning agents. This transition makes their ability to use external tools a critical capability. However, evaluating this skill presents a significant challenge. Existing benchmarks are often limited by their reliance on synthetic tools and severely constrained action spaces. To address these limitations, we introduce MCPVerse, an expansive, real-world benchmark for evaluating agentic tool use. MCPVerse integrates more than 550 real-world, executable tools to create an unprecedented action space exceeding 140k tokens, and employs outcome-based evaluation with real-time ground truth for time-sensitive tasks. We benchmarked the state-of-the-art LLMs across three modes (Oracle, Standard, and Max-Scale), revealing that while most models suffer performance degradation when confronted with larger tool sets, the agentic models, such as Claude-4-Sonnet, can effectively leverage expanded exploration spaces to improve accuracy. This finding not only exposes the limitations of state-of-the-art models in complex, real-world scenarios but also establishes MCPVerse as a critical benchmark for measuring and advancing agentic tool use capabilities.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-fbe907bf4cf5ecac", "title": "Limit-Computable Grains of Truth for Arbitrary Computable Extensive-Form (Un)Known Games", "authors": ["Cole Wyeth", "Marcus Hutter", "Jan Leike", "Jessica Taylor"], "year": 2025, "published": "2025-08-22T09:24:55Z", "updated": "2025-08-22T09:24:55Z", "arxiv_id": "http://arxiv.org/abs/2508.16245v1", "url": "https://arxiv.org/abs/2508.16245v1", "primary_category": "cs.GT", "categories": ["cs.GT", "cs.LG", "cs.MA", "econ.TH"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Deep Research"], "benchmarks": [], "models": [], "key_results": ["A Bayesian player acting in an infinite multi-player game learns to predict the other players' strategies if his prior assigns positive probability to their play (or contains a grain of truth)."], "abstract": "A Bayesian player acting in an infinite multi-player game learns to predict the other players' strategies if his prior assigns positive probability to their play (or contains a grain of truth). Kalai and Lehrer's classic grain of truth problem is to find a reasonably large class of strategies that contains the Bayes-optimal policies with respect to this class, allowing mutually-consistent beliefs about strategy choice that obey the rules of Bayesian inference. Only small classes are known to have a grain of truth and the literature contains several related impossibility results. In this paper we present a formal and general solution to the full grain of truth problem: we construct a class of strategies wide enough to contain all computable strategies as well as Bayes-optimal strategies for every reasonable prior over the class. When the \"environment\" is a known repeated stage game, we show convergence in the sense of [KL93a] and [KL93b]. When the environment is unknown, agents using Thompson sampling converge to play $\\varepsilon$-Nash equilibria in arbitrary unknown computable multi-agent environments. Finally, we include an application to self-predictive policies that avoid planning. While these results use computability theory only as a conceptual tool to solve a classic game theory problem, we show that our solution can naturally be computationally approximated arbitrarily closely.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-98c421f56b232d63", "title": "Graph RAG as Human Choice Model: Building a Data-Driven Mobility Agent with Preference Chain", "authors": ["Kai Hu", "Parfait Atchade-Adelomou", "Carlo Adornetto", "Adrian Mora-Carrero", "Luis Alonso-Pastor", "Ariel Noyman", "Yubo Liu", "Kent Larson"], "year": 2025, "published": "2025-08-22T07:50:57Z", "updated": "2025-09-05T08:26:57Z", "arxiv_id": "http://arxiv.org/abs/2508.16172v2", "url": "https://arxiv.org/abs/2508.16172v2", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["Experiments conducted on the Replica dataset demonstrate that the Preference Chain outperforms standard LLM in aligning with real-world transportation mode choices."], "abstract": "Understanding human behavior in urban environments is a crucial field within city sciences. However, collecting accurate behavioral data, particularly in newly developed areas, poses significant challenges. Recent advances in generative agents, powered by Large Language Models (LLMs), have shown promise in simulating human behaviors without relying on extensive datasets. Nevertheless, these methods often struggle with generating consistent, context-sensitive, and realistic behavioral outputs. To address these limitations, this paper introduces the Preference Chain, a novel method that integrates Graph Retrieval-Augmented Generation (RAG) with LLMs to enhance context-aware simulation of human behavior in transportation systems. Experiments conducted on the Replica dataset demonstrate that the Preference Chain outperforms standard LLM in aligning with real-world transportation mode choices. The development of the Mobility Agent highlights potential applications of proposed method in urban mobility modeling for emerging cities, personalized travel behavior analysis, and dynamic traffic forecasting. Despite limitations such as slow inference and the risk of hallucination, the method offers a promising framework for simulating complex human behavior in data-scarce environments, where traditional data-driven models struggle due to limited data availability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-a7b6cf47b8566603", "title": "Memento: Fine-tuning LLM Agents without Fine-tuning LLMs", "authors": ["Huichi Zhou", "Yihang Chen", "Siyuan Guo", "Xue Yan", "Kin Hei Lee", "Zihan Wang", "Ka Yiu Lee", "Guchun Zhang", "Kun Shao", "Linyi Yang", "Jun Wang"], "year": 2025, "published": "2025-08-22T07:25:30Z", "updated": "2025-08-25T13:32:12Z", "arxiv_id": "http://arxiv.org/abs/2508.16153v2", "url": "https://arxiv.org/abs/2508.16153v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Memory / RAG", "Deep Research"], "benchmarks": ["GAIA"], "models": ["ying"], "key_results": ["The policy is continually updated based on environmental feedback through a memory rewriting mechanism, whereas policy improvement is achieved through efficient memory reading (retrieval).", "It reaches $66.6\\%$ F1 and $80.4\\%$ PM on the DeepResearcher dataset, outperforming the state-of-the-art training-based method, while case-based memory adds $4.7\\%$ to $9.6\\%$ absolute points on out-of-distribution tasks."], "abstract": "In this paper, we introduce a novel learning paradigm for Adaptive Large Language Model (LLM) agents that eliminates the need for fine-tuning the underlying LLMs. Existing approaches are often either rigid, relying on static, handcrafted reflection workflows, or computationally intensive, requiring gradient updates of LLM model parameters. In contrast, our method enables low-cost continual adaptation via memory-based online reinforcement learning. We formalise this as a Memory-augmented Markov Decision Process (M-MDP), equipped with a neural case-selection policy to guide action decisions. Past experiences are stored in an episodic memory, either differentiable or non-parametric. The policy is continually updated based on environmental feedback through a memory rewriting mechanism, whereas policy improvement is achieved through efficient memory reading (retrieval). We instantiate our agent model in the deep research setting, namely \\emph{Memento}, which attains top-1 on GAIA validation ($87.88\\%$ Pass@$3$) and $79.40\\%$ on the test set. It reaches $66.6\\%$ F1 and $80.4\\%$ PM on the DeepResearcher dataset, outperforming the state-of-the-art training-based method, while case-based memory adds $4.7\\%$ to $9.6\\%$ absolute points on out-of-distribution tasks. Our approach offers a scalable and efficient pathway for developing generalist LLM agents capable of continuous, real-time learning without gradient updates, advancing machine learning towards open-ended skill acquisition and deep research scenarios. The code is available at https://github.com/Agent-on-the-Fly/Memento.", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1491a011ded5aa69", "title": "Anchor-MoE: A Mean-Anchored Mixture of Experts For Probabilistic Regression", "authors": ["Baozhuo Su", "Zhengxian Qu"], "year": 2025, "published": "2025-08-22T21:12:41Z", "updated": "2025-08-22T21:12:41Z", "arxiv_id": "http://arxiv.org/abs/2508.16802v1", "url": "https://arxiv.org/abs/2508.16802v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["MoE"], "benchmarks": [], "models": [], "key_results": ["For simplicity, we use a tuned gradient-boosting model to furnish the anchor mean; however, any off-the-shelf point regressor can serve as the anchor.", "We train by minimizing negative log-likelihood, and on a disjoint calibration split fit a post-hoc linear map on predicted means to improve point accuracy.", "Empirically, across standard UCI regressions, Anchor-MoE consistently matches or surpasses the strong NGBoost baseline in RMSE and NLL; on several datasets it achieves new state-of-the-art probabilistic regression results on our benchmark suite."], "abstract": "Regression under uncertainty is fundamental across science and engineering. We present an Anchored Mixture of Experts (Anchor-MoE), a model that handles both probabilistic and point regression. For simplicity, we use a tuned gradient-boosting model to furnish the anchor mean; however, any off-the-shelf point regressor can serve as the anchor. The anchor prediction is projected into a latent space, where a learnable metric-window kernel scores locality and a soft router dispatches each sample to a small set of mixture-density-network experts; the experts produce a heteroscedastic correction and predictive variance. We train by minimizing negative log-likelihood, and on a disjoint calibration split fit a post-hoc linear map on predicted means to improve point accuracy. On the theory side, assuming a Hölder smooth regression function of order~$α$ and fixed Lipschitz partition-of-unity weights with bounded overlap, we show that Anchor-MoE attains the minimax-optimal $L^2$ risk rate $O\\!\\big(N^{-2α/(2α+d)}\\big)$. In addition, the CRPS test generalization gap scales as $\\widetilde{O}\\!\\Big(\\sqrt{(\\log(Mh)+P+K)/N}\\Big)$; it is logarithmic in $Mh$ and scales as the square root in $P$ and $K$. Under bounded-overlap routing, $K$ can be replaced by $k$, and any dependence on a latent dimension is absorbed into $P$. Under uniformly bounded means and variances, an analogous $\\widetilde{O}\\!\\big(\\sqrt{(\\log(Mh)+P+K)/N}\\big)$ scaling holds for the test NLL up to constants. Empirically, across standard UCI regressions, Anchor-MoE consistently matches or surpasses the strong NGBoost baseline in RMSE and NLL; on several datasets it achieves new state-of-the-art probabilistic regression results on our benchmark suite. Code is available at https://github.com/BaozhuoSU/Probabilistic_Regression.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-99006348aaa7148a", "title": "TaDiCodec: Text-aware Diffusion Speech Tokenizer for Speech Language Modeling", "authors": ["Yuancheng Wang", "Dekun Chen", "Xueyao Zhang", "Junan Zhang", "Jiaqi Li", "Zhizheng Wu"], "year": 2025, "published": "2025-08-22T20:45:03Z", "updated": "2025-08-22T20:45:03Z", "arxiv_id": "http://arxiv.org/abs/2508.16790v1", "url": "https://arxiv.org/abs/2508.16790v1", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.LG", "eess.AS"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Quantization", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["TaDiCodec employs end-to-end optimization for quantization and reconstruction through a diffusion autoencoder, while integrating text guidance into the diffusion decoder to enhance reconstruction quality and achieve optimal compression.", "TaDiCodec achieves an extremely low frame rate of 6.25 Hz and a corresponding bitrate of 0.0875 kbps with a single-layer codebook for 24 kHz speech, while maintaining superior performance on critical speech generation evaluation metrics such as Word Error Rate (WER), speaker similarity (SIM), and speech quality (UTMOS)."], "abstract": "Speech tokenizers serve as foundational components for speech language models, yet current designs exhibit several limitations, including: 1) dependence on multi-layer residual vector quantization structures or high frame rates, 2) reliance on auxiliary pre-trained models for semantic distillation, and 3) requirements for complex two-stage training processes. In this work, we introduce the Text-aware Diffusion Transformer Speech Codec (TaDiCodec), a novel approach designed to overcome these challenges. TaDiCodec employs end-to-end optimization for quantization and reconstruction through a diffusion autoencoder, while integrating text guidance into the diffusion decoder to enhance reconstruction quality and achieve optimal compression. TaDiCodec achieves an extremely low frame rate of 6.25 Hz and a corresponding bitrate of 0.0875 kbps with a single-layer codebook for 24 kHz speech, while maintaining superior performance on critical speech generation evaluation metrics such as Word Error Rate (WER), speaker similarity (SIM), and speech quality (UTMOS). Notably, TaDiCodec employs a single-stage, end-to-end training paradigm, and obviating the need for auxiliary pre-trained models. We also validate the compatibility of TaDiCodec in language model based zero-shot text-to-speech with both autoregressive modeling and masked generative modeling, demonstrating its effectiveness and efficiency for speech language modeling, as well as a significantly small reconstruction-generation gap. We will open source our code and model checkpoints. Audio samples are are available at https:/tadicodec.github.io/. We release code and model checkpoints at https:/github.com/HeCheng0625/Diffusion-Speech-Tokenizer.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-19b7a3e5e23ce946", "title": "Interpreting the Effects of Quantization on LLMs", "authors": ["Manpreet Singh", "Hassan Sajjad"], "year": 2025, "published": "2025-08-22T20:36:53Z", "updated": "2025-11-20T02:20:31Z", "arxiv_id": "http://arxiv.org/abs/2508.16785v3", "url": "https://arxiv.org/abs/2508.16785v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Interpretability"], "benchmarks": [], "models": ["Llama-2-7B."], "key_results": ["Quantization offers a practical solution to deploy LLMs in resource-constraint environments."], "abstract": "Quantization offers a practical solution to deploy LLMs in resource-constraint environments. However, its impact on internal representations remains understudied, raising questions about the reliability of quantized models. In this study, we employ a range of interpretability techniques to investigate how quantization affects model and neuron behavior. We analyze multiple LLMs under 4-bit and 8-bit quantization. Our findings reveal that the impact of quantization on model calibration is generally minor. Analysis of neuron activations indicates that the number of dead neurons, i.e., those with activation values close to 0 across the dataset, remains consistent regardless of quantization. In terms of neuron contribution to predictions, we observe that smaller full precision models exhibit fewer salient neurons, whereas larger models tend to have more, with the exception of Llama-2-7B. The effect of quantization on neuron redundancy varies across models. Overall, our findings suggest that effect of quantization may vary by model and tasks, however, we did not observe any drastic change which may discourage the use of quantization as a reliable model compression technique.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1a53fcb96e1d5667", "title": "Closer to Reality: Practical Semi-Supervised Federated Learning for Foundation Model Adaptation", "authors": ["Guangyu Sun", "Jingtao Li", "Weiming Zhuang", "Chen Chen", "Chen Chen", "Lingjuan Lyu"], "year": 2025, "published": "2025-08-22T17:47:02Z", "updated": "2025-08-22T17:47:02Z", "arxiv_id": "http://arxiv.org/abs/2508.16568v1", "url": "https://arxiv.org/abs/2508.16568v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "Memory / RAG"], "benchmarks": [], "models": ["ying"], "key_results": ["Foundation models (FMs) exhibit remarkable generalization but require adaptation to downstream tasks, particularly in privacy-sensitive applications."], "abstract": "Foundation models (FMs) exhibit remarkable generalization but require adaptation to downstream tasks, particularly in privacy-sensitive applications. Due to data privacy regulations, cloud-based FMs cannot directly access private edge data, limiting their adaptation. Federated learning (FL) provides a privacy-aware alternative, but existing FL approaches overlook the constraints imposed by edge devices -- namely, limited computational resources and the scarcity of labeled data. To address these challenges, we introduce Practical Semi-Supervised Federated Learning (PSSFL), where edge devices hold only unlabeled, low-resolution data, while the server has limited labeled, high-resolution data. In this setting, we propose the Federated Mixture of Experts (FedMox), a novel framework that enhances FM adaptation in FL. FedMox tackles computational and resolution mismatch challenges via a sparse Mixture-of-Experts architecture, employing a spatial router to align features across resolutions and a Soft-Mixture strategy to stabilize semi-supervised learning. We take object detection as a case study, and experiments on real-world autonomous driving datasets demonstrate that FedMox effectively adapts FMs under PSSFL, significantly improving performance with constrained memory costs on edge devices. Our work paves the way for scalable and privacy-preserving FM adaptation in federated scenarios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7bb5b101f0380fdb", "title": "Parameter-Free Logit Distillation via Sorting Mechanism", "authors": ["Stephen Ekaputra Limantoro"], "year": 2025, "published": "2025-08-22T17:09:38Z", "updated": "2025-08-22T17:09:38Z", "arxiv_id": "http://arxiv.org/abs/2508.16544v1", "url": "https://arxiv.org/abs/2508.16544v1", "primary_category": "eess.SP", "categories": ["eess.SP", "cs.LG", "eess.IV"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": ["ImageNet"], "models": [], "key_results": ["Knowledge distillation (KD) aims to distill the knowledge from the teacher (larger) to the student (smaller) model via soft-label for the efficient neural network."], "abstract": "Knowledge distillation (KD) aims to distill the knowledge from the teacher (larger) to the student (smaller) model via soft-label for the efficient neural network. In general, the performance of a model is determined by accuracy, which is measured with labels. However, existing KD approaches usually use the teacher with its original distribution, neglecting the potential of incorrect prediction. This may contradict the motivation of hard-label learning through cross-entropy loss, which may lead to sub-optimal knowledge distillation on certain samples. To address this issue, we propose a novel logit processing scheme via a sorting mechanism. Specifically, our method has a two-fold goal: (1) fixing the incorrect prediction of the teacher based on the labels and (2) reordering the distribution in a natural way according to priority rank at once. As an easy-to-use, plug-and-play pre-processing, our sort method can be effectively applied to existing logit-based KD methods. Extensive experiments on the CIFAR-100 and ImageNet datasets demonstrate the effectiveness of our method.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d851c88a61c357be", "title": "Systematic Characterization of LLM Quantization: A Performance, Energy, and Quality Perspective", "authors": ["Tianyao Shi", "Yi Ding"], "year": 2025, "published": "2025-08-22T14:59:23Z", "updated": "2025-08-22T14:59:23Z", "arxiv_id": "http://arxiv.org/abs/2508.16712v1", "url": "https://arxiv.org/abs/2508.16712v1", "primary_category": "cs.PF", "categories": ["cs.PF", "cs.AI", "cs.AR", "cs.DC", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["Large language models (LLMs) have demonstrated remarkable capabilities across diverse domains, but their heavy resource demands make quantization-reducing precision to lower-bit formats-critical for efficient serving."], "abstract": "Large language models (LLMs) have demonstrated remarkable capabilities across diverse domains, but their heavy resource demands make quantization-reducing precision to lower-bit formats-critical for efficient serving. While many quantization methods exist, a systematic understanding of their performance, energy, and quality tradeoffs in realistic serving conditions remains a gap. In this work, we first develop a fully automated online characterization framework qMeter, and then conduct an in-depth characterization of 11 post-training LLM quantization methods across 4 model sizes (7B-70B) and two GPU architectures (A100, H100). We evaluate quantization at the application, workload, parallelism, and hardware levels under online serving conditions. Our study reveals highly task- and method-dependent tradeoffs, strong sensitivity to workload characteristics, and complex interactions with parallelism and GPU architecture. We further present three optimization case studies illustrating deployment challenges in capacity planning, energy-efficient scheduling, and multi-objective tuning. To the best of our knowledge, this is one of the first comprehensive application-, system-, and hardware-level characterization of LLM quantization from a joint performance, energy, and quality perspective.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1abb813e151e4867", "title": "Sparse and Dense Retrievers Learn Better Together: Joint Sparse-Dense Optimization for Text-Image Retrieval", "authors": ["Jonghyun Song", "Youngjune Lee", "Gyu-Hwung Cho", "Ilhyeon Song", "Saehun Kim", "Yohan Jo"], "year": 2025, "published": "2025-08-22T13:25:58Z", "updated": "2025-08-22T13:25:58Z", "arxiv_id": "http://arxiv.org/abs/2508.16707v1", "url": "https://arxiv.org/abs/2508.16707v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.IR", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Interpretability", "VLM"], "benchmarks": [], "models": [], "key_results": ["Vision-Language Pretrained (VLP) models have achieved impressive performance on multimodal tasks, including text-image retrieval, based on dense representations.", "This bi-directional learning is achieved using an integrated similarity score-a weighted sum of dense and sparse similarities-which serves as a shared teacher signal for both representations.", "Experiments on MSCOCO and Flickr30k demonstrate that our sparse retriever not only outperforms existing sparse baselines, but also achieves performance comparable to-or even surpassing-its dense counterparts, while retaining the benefits of sparse models."], "abstract": "Vision-Language Pretrained (VLP) models have achieved impressive performance on multimodal tasks, including text-image retrieval, based on dense representations. Meanwhile, Learned Sparse Retrieval (LSR) has gained traction in text-only settings due to its interpretability and efficiency with fast term-based lookup via inverted indexes. Inspired by these advantages, recent work has extended LSR to the multimodal domain. However, these methods often rely on computationally expensive contrastive pre-training, or distillation from a frozen dense model, which limits the potential for mutual enhancement. To address these limitations, we propose a simple yet effective framework that enables bi-directional learning between dense and sparse representations through Self-Knowledge Distillation. This bi-directional learning is achieved using an integrated similarity score-a weighted sum of dense and sparse similarities-which serves as a shared teacher signal for both representations. To ensure efficiency, we fine-tune the final layer of the dense encoder and the sparse projection head, enabling easy adaptation of any existing VLP model. Experiments on MSCOCO and Flickr30k demonstrate that our sparse retriever not only outperforms existing sparse baselines, but also achieves performance comparable to-or even surpassing-its dense counterparts, while retaining the benefits of sparse models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0dc917528d6e14b2", "title": "SpecVLM: Enhancing Speculative Decoding of Video LLMs via Verifier-Guided Token Pruning", "authors": ["Yicheng Ji", "Jun Zhang", "Heming Xia", "Jinpeng Chen", "Lidan Shou", "Gang Chen", "Huan Li"], "year": 2025, "published": "2025-08-22T08:23:09Z", "updated": "2025-08-28T06:44:28Z", "arxiv_id": "http://arxiv.org/abs/2508.16201v2", "url": "https://arxiv.org/abs/2508.16201v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Verifier / PRM", "Speculative Decoding", "Memory / RAG"], "benchmarks": [], "models": ["Qwen2.5-VL-32B.", "LLaVA-OneVision-72B", "yicheng"], "key_results": ["To achieve this, we performs a two-stage pruning process: Stage I selects highly informative tokens guided by attention signals from the verifier (target model), while Stage II prunes remaining redundant ones in a spatially uniform manner.", "Extensive experiments on four video understanding benchmarks demonstrate the effectiveness and robustness of SpecVLM, which achieves up to 2.68$\\times$ decoding speedup for LLaVA-OneVision-72B and 2.11$\\times$ speedup for Qwen2.5-VL-32B."], "abstract": "Video large language models (Vid-LLMs) have shown strong capabilities in understanding video content. However, their reliance on dense video token representations introduces substantial memory and computational overhead in both prefilling and decoding. To mitigate the information loss of recent video token reduction methods and accelerate the decoding stage of Vid-LLMs losslessly, we introduce SpecVLM, a training-free speculative decoding (SD) framework tailored for Vid-LLMs that incorporates staged video token pruning. Building on our novel finding that the draft model's speculation exhibits low sensitivity to video token pruning, SpecVLM prunes up to 90% of video tokens to enable efficient speculation without sacrificing accuracy. To achieve this, we performs a two-stage pruning process: Stage I selects highly informative tokens guided by attention signals from the verifier (target model), while Stage II prunes remaining redundant ones in a spatially uniform manner. Extensive experiments on four video understanding benchmarks demonstrate the effectiveness and robustness of SpecVLM, which achieves up to 2.68$\\times$ decoding speedup for LLaVA-OneVision-72B and 2.11$\\times$ speedup for Qwen2.5-VL-32B. Code is available at https://github.com/zju-jiyicheng/SpecVLM.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-2c9e1f8955d783f8", "title": "ShadowNPU: System and Algorithm Co-design for NPU-Centric On-Device LLM Inference", "authors": ["Wangsong Yin", "Daliang Xu", "Mengwei Xu", "Gang Huang", "Xuanzhe Liu"], "year": 2025, "published": "2025-08-22T07:41:35Z", "updated": "2026-04-08T06:56:34Z", "arxiv_id": "http://arxiv.org/abs/2508.16703v4", "url": "https://arxiv.org/abs/2508.16703v4", "primary_category": "cs.PF", "categories": ["cs.PF", "cs.AI", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["We observe that the attention operator falls back from the special-purpose NPU to the general-purpose CPU/GPU because of quantization sensitivity in state-of-the-art frameworks.", "Further, shadowAttn proposes insightful techniques such as NPU compute graph bucketing, head-wise NPU-CPU/GPU pipeline and per-head fine-grained sparsity ratio to achieve high accuracy and efficiency.", "shadowAttn delivers the best performance with highly limited CPU/GPU resource; it requires much less CPU/GPU resource to deliver on-par performance of SoTA frameworks."], "abstract": "On-device running Large Language Models (LLMs) is nowadays a critical enabler towards preserving user privacy. We observe that the attention operator falls back from the special-purpose NPU to the general-purpose CPU/GPU because of quantization sensitivity in state-of-the-art frameworks. This fallback results in a degraded user experience and increased complexity in system scheduling. To this end, this paper presents shadowAttn, a system-algorithm codesigned sparse attention module with minimal reliance on CPU/GPU by only sparsely calculating the attention on a tiny portion of tokens. The key idea is to hide the overhead of estimating the important tokens with a NPU-based pilot compute. Further, shadowAttn proposes insightful techniques such as NPU compute graph bucketing, head-wise NPU-CPU/GPU pipeline and per-head fine-grained sparsity ratio to achieve high accuracy and efficiency. shadowAttn delivers the best performance with highly limited CPU/GPU resource; it requires much less CPU/GPU resource to deliver on-par performance of SoTA frameworks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6844aa4cb682fbc3", "title": "CommonKV: Compressing KV Cache with Cross-layer Parameter Sharing", "authors": ["Yixuan Wang", "Haoyu Qiao", "Lujun Li", "Qingfu Zhu", "Wanxiang Che"], "year": 2025, "published": "2025-08-22T06:55:45Z", "updated": "2025-08-22T06:55:45Z", "arxiv_id": "http://arxiv.org/abs/2508.16134v1", "url": "https://arxiv.org/abs/2508.16134v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "KV Cache", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Inspired by the high similarity observed in cross-layer hidden states, we utilize Singular Value Decomposition (SVD) to achieve weight sharing across adjacent parameters, resulting in a more easily mergeable latent KV cache.", "Experiments across multiple backbone models and benchmarks including LongBench and Ruler demonstrate that the proposed method consistently outperforms existing low-rank and cross-layer approaches at various compression ratios.", "By integrating these approaches, we can ultimately achieve a 98\\% compression ratio without significant performance loss."], "abstract": "Large Language Models (LLMs) confront significant memory challenges due to the escalating KV cache with increasing sequence length. As a crucial technique, existing cross-layer KV cache sharing methods either necessitate modified model architectures with subsequent pre-training or incur significant performance degradation at high compression rates. To mitigate these challenges, we propose CommonKV, a training-free method for cross-layer KV cache compression through adjacent parameters sharing. Inspired by the high similarity observed in cross-layer hidden states, we utilize Singular Value Decomposition (SVD) to achieve weight sharing across adjacent parameters, resulting in a more easily mergeable latent KV cache. Furthermore, we also introduce an adaptive budget allocation strategy. It dynamically assigns compression budgets based on cosine similarity, ensuring that dissimilar caches are not over-compressed. Experiments across multiple backbone models and benchmarks including LongBench and Ruler demonstrate that the proposed method consistently outperforms existing low-rank and cross-layer approaches at various compression ratios. Moreover, we find that the benefits of CommonKV are orthogonal to other quantization and eviction methods. By integrating these approaches, we can ultimately achieve a 98\\% compression ratio without significant performance loss.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-107655d27baba6ca", "title": "Domain Adaptation via Feature Refinement", "authors": ["Savvas Karatsiolis", "Andreas Kamilaris"], "year": 2025, "published": "2025-08-22T06:32:19Z", "updated": "2025-08-22T06:32:19Z", "arxiv_id": "http://arxiv.org/abs/2508.16124v1", "url": "https://arxiv.org/abs/2508.16124v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Preference Optimization"], "benchmarks": [], "models": ["R10-C", "R100-C", "phisticated"], "key_results": ["Extensive experiments on benchmark datasets, including CIFAR10-C, CIFAR100-C, MNIST-C and PatchCamelyon-C, demonstrate that the proposed algorithm outperforms prior methods in robustness to corruption.", "Theoretical and empirical analyses further reveal that our method achieves improved feature alignment, increased mutual information between the domains and reduced sensitivity to input perturbations."], "abstract": "We propose Domain Adaptation via Feature Refinement (DAFR2), a simple yet effective framework for unsupervised domain adaptation under distribution shift. The proposed method synergistically combines three key components: adaptation of Batch Normalization statistics using unlabeled target data, feature distillation from a source-trained model and hypothesis transfer. By aligning feature distributions at the statistical and representational levels, DAFR2 produces robust and domain-invariant feature spaces that generalize across similar domains without requiring target labels, complex architectures or sophisticated training objectives. Extensive experiments on benchmark datasets, including CIFAR10-C, CIFAR100-C, MNIST-C and PatchCamelyon-C, demonstrate that the proposed algorithm outperforms prior methods in robustness to corruption. Theoretical and empirical analyses further reveal that our method achieves improved feature alignment, increased mutual information between the domains and reduced sensitivity to input perturbations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-36c2bf157ffa94d5", "title": "MultiPL-MoE: Multi-Programming-Lingual Extension of Large Language Models through Hybrid Mixture-of-Experts", "authors": ["Qing Wang", "Xue Han", "Jiahui Wang", "Lehao Xing", "Qian Hu", "Lianlian Zhang", "Chao Deng", "Junlan Feng"], "year": 2025, "published": "2025-08-22T06:24:52Z", "updated": "2025-09-08T08:30:07Z", "arxiv_id": "http://arxiv.org/abs/2508.19268v2", "url": "https://arxiv.org/abs/2508.19268v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["MoE"], "benchmarks": [], "models": [], "key_results": ["To address this, we intent to improve the multi-programming-lingual (MultiPL) performance of the base LLMs while retaining the most popular ones using restricted computational resources."], "abstract": "Despite LLMs' excellent code creation capabilities, multilingual code generation remains extremely challenging. To address this, we intent to improve the multi-programming-lingual (MultiPL) performance of the base LLMs while retaining the most popular ones using restricted computational resources. We consider MultiPL to be a special case of multiple natural languages and propose a MultiPL extension of LLMs utilizing a hybrid mixture of experts (MoE), called MultiPL-MoE. Specifically, MultiPL-MoE combines two paired MoEs to optimize expert selection at both the token and segment levels. The token-level MoE is a standard upcycling MoE structure with a shared expert and a novel gate weight normalization approach that aids in the final fusion with the segment-level MoE. The segment-level MoE incorporates two innovative designs to better capture the syntactic structure and contextual patterns of programming languages: First, using a sliding window to partition the input token sequence into multiple segments; Then, adopting an expert-choice routing strategy that allows experts to select the top-k segments. The results of the experiment proved the effectiveness of MultiPL-MoE.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-3ed7d9fd7454dd1a", "title": "GPT-OSS-20B: A Comprehensive Deployment-Centric Analysis of OpenAI's Open-Weight Mixture of Experts Model", "authors": ["Deepak Kumar", "Divakar Yadav", "Yash Patel"], "year": 2025, "published": "2025-08-22T03:37:27Z", "updated": "2025-08-31T03:40:19Z", "arxiv_id": "http://arxiv.org/abs/2508.16700v2", "url": "https://arxiv.org/abs/2508.16700v2", "primary_category": "cs.AR", "categories": ["cs.AR", "cs.AI", "cs.DC", "cs.PF"], "pillar": "Efficient Training & Inference", "methods": ["MoE"], "benchmarks": [], "models": ["Qwen3-32B", "Yi-34B"], "key_results": ["3.61B active) against dense baselines Qwen3-32B and Yi-34B across multiple dimensions."], "abstract": "We present a single-GPU (H100, bf16) evaluation of GPT-OSS-20B (Mixture-of-Experts; 20.9B total, approx. 3.61B active) against dense baselines Qwen3-32B and Yi-34B across multiple dimensions. We measure true time-to-first-token (TTFT), full-decode throughput (TPOT), end-to-end latency percentiles, peak VRAM with past key values (PKV) held, and energy via a consistent nvidia-smi-based sampler. At a 2048-token context with 64-token decode, GPT-OSS-20B delivers higher decode throughput and tokens per Joule than dense baselines Qwen3-32B and Yi-34B, while substantially reducing peak VRAM and energy per 1000 generated tokens; its TTFT is higher due to MoE routing overhead. With only 17.3% of parameters active (3.61B of 20.9B), GPT-OSS-20B provides about 31.8% higher decode throughput and 25.8% lower energy per 1000 generated tokens than Qwen3-32B at 2048/64, while using 31.7% less peak VRAM. Normalized by active parameters, GPT-OSS-20B shows markedly stronger per-active-parameter efficiency (APE), underscoring MoE's deployment advantages. We do not evaluate accuracy; this is a deployment-focused study. We release code and consolidated results to enable replication and extension.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-99a919208a4075e2", "title": "Pareto Actor-Critic for Communication and Computation Co-Optimization in Non-Cooperative Federated Learning Services", "authors": ["Renxuan Tan", "Rongpeng Li", "Xiaoxue Yu", "Xianfu Chen", "Xing Xu", "Zhifeng Zhao"], "year": 2025, "published": "2025-08-22T02:09:48Z", "updated": "2025-08-28T05:26:41Z", "arxiv_id": "http://arxiv.org/abs/2508.16037v2", "url": "https://arxiv.org/abs/2508.16037v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "Quantization", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Within the framework, we integrate Pareto Actor-Critic (PAC) principles with expectile regression, enabling agents to conjecture optimal joint policies to achieve Pareto-optimal equilibria while modeling heterogeneous risk profiles.", "Further, we develop PAC-MCoFL-p, a scalable variant featuring a parameterized conjecture generator that substantially reduces computational complexity with a provably bounded error.", "Alongside theoretical convergence guarantees, our framework's superiority is validated through extensive simulations -- PAC-MCoFL achieves approximately 5.8% and 4.2% improvements in total reward and hypervolume indicator (HVI), respectively, over the latest MARL solutions."], "abstract": "Federated learning (FL) in multi-service provider (SP) ecosystems is fundamentally hampered by non-cooperative dynamics, where privacy constraints and competing interests preclude the centralized optimization of multi-SP communication and computation resources. In this paper, we introduce PAC-MCoFL, a game-theoretic multi-agent reinforcement learning (MARL) framework where SPs act as agents to jointly optimize client assignment, adaptive quantization, and resource allocation. Within the framework, we integrate Pareto Actor-Critic (PAC) principles with expectile regression, enabling agents to conjecture optimal joint policies to achieve Pareto-optimal equilibria while modeling heterogeneous risk profiles. To manage the high-dimensional action space, we devise a ternary Cartesian decomposition (TCAD) mechanism that facilitates fine-grained control. Further, we develop PAC-MCoFL-p, a scalable variant featuring a parameterized conjecture generator that substantially reduces computational complexity with a provably bounded error. Alongside theoretical convergence guarantees, our framework's superiority is validated through extensive simulations -- PAC-MCoFL achieves approximately 5.8% and 4.2% improvements in total reward and hypervolume indicator (HVI), respectively, over the latest MARL solutions. The results also demonstrate that our method can more effectively balance individual SP and system performance in scaled deployments and under diverse data heterogeneity.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-25a5ba9deb691b17", "title": "Do Cognitively Interpretable Reasoning Traces Improve LLM Performance?", "authors": ["Siddhant Bhambri", "Upasana Biswas", "Subbarao Kambhampati"], "year": 2025, "published": "2025-08-21T23:48:50Z", "updated": "2025-08-21T23:48:50Z", "arxiv_id": "http://arxiv.org/abs/2508.16695v1", "url": "https://arxiv.org/abs/2508.16695v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Chain-of-Thought", "Distillation", "Interpretability"], "benchmarks": [], "models": ["LLaMA and", "Qwen", "DeepSeek", "R1", "yields"], "key_results": ["Recent progress in reasoning-oriented Large Language Models (LLMs) has been driven by introducing Chain-of-Thought (CoT) traces, where models generate intermediate reasoning traces before producing an answer."], "abstract": "Recent progress in reasoning-oriented Large Language Models (LLMs) has been driven by introducing Chain-of-Thought (CoT) traces, where models generate intermediate reasoning traces before producing an answer. These traces, as in DeepSeek R1, are not only used to guide inference but also serve as supervision signals for distillation into smaller models. A common but often implicit assumption is that CoT traces should be semantically meaningful and interpretable to the end user. While recent research questions the need for semantic nature of these traces, in this paper, we ask: ``\\textit{Must CoT reasoning traces be interpretable to enhance LLM task performance?}\" We investigate this question in the Open Book Question-Answering domain by supervised fine-tuning LLaMA and Qwen models on four types of reasoning traces: (1) DeepSeek R1 traces, (2) LLM-generated summaries of R1 traces, (3) LLM-generated post-hoc explanations of R1 traces, and (4) algorithmically generated verifiably correct traces. To quantify the trade-off between interpretability and performance, we further conduct a human-subject study with 100 participants rating the interpretability of each trace type. Our results reveal a striking mismatch: while fine-tuning on R1 traces yields the strongest performance, participants judged these traces to be the least interpretable. These findings suggest that it is useful to decouple intermediate tokens from end user interpretability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-3f3e0670b97e9656", "title": "Scalable Equilibrium Propagation via Intermediate Error Signals for Deep Convolutional CRNNs", "authors": ["Jiaqi Lin", "Malyaban Bal", "Abhronil Sengupta"], "year": 2025, "published": "2025-08-21T22:19:30Z", "updated": "2026-05-08T16:14:11Z", "arxiv_id": "http://arxiv.org/abs/2508.15989v2", "url": "https://arxiv.org/abs/2508.15989v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.ET"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["phic"], "key_results": ["Our proposed approach achieves state-of-the-art performance on the CIFAR-10 and CIFAR-100 datasets, showcasing its scalability on deep VGG architectures."], "abstract": "Equilibrium Propagation (EP) is a biologically inspired local learning rule first proposed for convergent recurrent neural networks (CRNNs), in which synaptic updates depend only on neuron states from two distinct phases. EP estimates gradients that closely align with those computed by Backpropagation Through Time (BPTT) while significantly reducing computational demands, positioning it as a potential candidate for on-chip training in neuromorphic architectures. However, prior studies on EP have been constrained to shallow architectures, as deeper networks suffer from the vanishing gradient problem, leading to convergence difficulties in both energy minimization and gradient computation. To alleviate the vanishing gradient problem in deep EP networks, we propose a novel EP framework that incorporates layer-wise learning signals to provide auxiliary supervision, which enhances the convergence of neuron dynamics. This is the first work to integrate knowledge distillation and local error signals into EP, enabling the training of significantly deeper architectures. Our proposed approach achieves state-of-the-art performance on the CIFAR-10 and CIFAR-100 datasets, showcasing its scalability on deep VGG architectures. These results represent a significant advancement in the scalability of EP, suggesting that intermediate learning signals can extend the practical applicability of EP to deeper architectures.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9f36818270503a13", "title": "Intern-S1: A Scientific Multimodal Foundation Model", "authors": ["Lei Bai", "Zhongrui Cai", "Yuhang Cao", "Maosong Cao", "Weihan Cao", "Chiyu Chen", "Haojiong Chen", "Kai Chen", "Pengcheng Chen", "Ying Chen", "Yongkang Chen", "Yu Cheng", "Pei Chu", "Tao Chu", "Erfei Cui", "Ganqu Cui", "Long Cui", "Ziyun Cui", "Nianchen Deng", "Ning Ding", "Nanqing Dong", "Peijie Dong", "Shihan Dou", "Sinan Du", "Haodong Duan", "Caihua Fan", "Ben Gao", "Changjiang Gao", "Jianfei Gao", "Songyang Gao", "Yang Gao", "Zhangwei Gao", "Jiaye Ge", "Qiming Ge", "Lixin Gu", "Yuzhe Gu", "Aijia Guo", "Qipeng Guo", "Xu Guo", "Conghui He", "Junjun He", "Yili Hong", "Siyuan Hou", "Caiyu Hu", "Hanglei Hu", "Jucheng Hu", "Ming Hu", "Zhouqi Hua", "Haian Huang", "Junhao Huang", "Xu Huang", "Zixian Huang", "Zhe Jiang", "Lingkai Kong", "Linyang Li", "Peiji Li", "Pengze Li", "Shuaibin Li", "Tianbin Li", "Wei Li", "Yuqiang Li", "Dahua Lin", "Junyao Lin", "Tianyi Lin", "Zhishan Lin", "Hongwei Liu", "Jiangning Liu", "Jiyao Liu", "Junnan Liu", "Kai Liu", "Kaiwen Liu", "Kuikun Liu", "Shichun Liu", "Shudong Liu", "Wei Liu", "Xinyao Liu", "Yuhong Liu", "Zhan Liu", "Yinquan Lu", "Haijun Lv", "Hongxia Lv", "Huijie Lv", "Qitan Lv", "Ying Lv", "Chengqi Lyu", "Chenglong Ma", "Jianpeng Ma", "Ren Ma", "Runmin Ma", "Runyuan Ma", "Xinzhu Ma", "Yichuan Ma", "Zihan Ma", "Sixuan Mi", "Junzhi Ning", "Wenchang Ning", "Xinle Pang", "Jiahui Peng", "Runyu Peng", "Yu Qiao", "Jiantao Qiu", "Xiaoye Qu", "Yuan Qu", "Yuchen Ren", "Fukai Shang", "Wenqi Shao", "Junhao Shen", "Shuaike Shen", "Chunfeng Song", "Demin Song", "Diping Song", "Chenlin Su", "Weijie Su", "Weigao Sun", "Yu Sun", "Qian Tan", "Cheng Tang", "Huanze Tang", "Kexian Tang", "Shixiang Tang", "Jian Tong", "Aoran Wang", "Bin Wang", "Dong Wang", "Lintao Wang", "Rui Wang", "Weiyun Wang", "Wenhai Wang", "Jiaqi Wang", "Yi Wang", "Ziyi Wang", "Ling-I Wu", "Wen Wu", "Yue Wu", "Zijian Wu", "Linchen Xiao", "Shuhao Xing", "Chao Xu", "Huihui Xu", "Jun Xu", "Ruiliang Xu", "Wanghan Xu", "GanLin Yang", "Yuming Yang", "Haochen Ye", "Jin Ye", "Shenglong Ye", "Jia Yu", "Jiashuo Yu", "Jing Yu", "Fei Yuan", "Yuhang Zang", "Bo Zhang", "Chao Zhang", "Chen Zhang", "Hongjie Zhang", "Jin Zhang", "Qiaosheng Zhang", "Qiuyinzhe Zhang", "Songyang Zhang", "Taolin Zhang", "Wenlong Zhang", "Wenwei Zhang", "Yechen Zhang", "Ziyang Zhang", "Haiteng Zhao", "Qian Zhao", "Xiangyu Zhao", "Xiangyu Zhao", "Bowen Zhou", "Dongzhan Zhou", "Peiheng Zhou", "Yuhao Zhou", "Yunhua Zhou", "Dongsheng Zhu", "Lin Zhu", "Yicheng Zou"], "year": 2025, "published": "2025-08-21T17:58:00Z", "updated": "2025-08-24T19:35:34Z", "arxiv_id": "http://arxiv.org/abs/2508.15763v2", "url": "https://arxiv.org/abs/2508.15763v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "MoE", "VLM"], "benchmarks": [], "models": [], "key_results": ["In recent years, a plethora of open-source foundation models have emerged, achieving remarkable progress in some widely attended fields, with performance being quite close to that of closed-source models.", "Through integrated innovations in algorithms, data, and training systems, Intern-S1 achieved top-tier performance in online RL training.", "On comprehensive evaluation benchmarks, Intern-S1 demonstrates competitive performance on general reasoning tasks among open-source models and significantly outperforms open-source models in scientific domains, surpassing closed-source state-of-the-art models in professional tasks, such as molecular synthesis planning, reaction condition prediction, predicting thermodynamic stabilities for crystals."], "abstract": "In recent years, a plethora of open-source foundation models have emerged, achieving remarkable progress in some widely attended fields, with performance being quite close to that of closed-source models. However, in high-value but more challenging scientific professional fields, either the fields still rely on expert models, or the progress of general foundation models lags significantly compared to those in popular areas, far from sufficient for transforming scientific research and leaving substantial gap between open-source models and closed-source models in these scientific domains. To mitigate this gap and explore a step further toward Artificial General Intelligence (AGI), we introduce Intern-S1, a specialized generalist equipped with general understanding and reasoning capabilities with expertise to analyze multiple science modal data. Intern-S1 is a multimodal Mixture-of-Experts (MoE) model with 28 billion activated parameters and 241 billion total parameters, continually pre-trained on 5T tokens, including over 2.5T tokens from scientific domains. In the post-training stage, Intern-S1 undergoes offline and then online reinforcement learning (RL) in InternBootCamp, where we propose Mixture-of-Rewards (MoR) to synergize the RL training on more than 1000 tasks simultaneously. Through integrated innovations in algorithms, data, and training systems, Intern-S1 achieved top-tier performance in online RL training. On comprehensive evaluation benchmarks, Intern-S1 demonstrates competitive performance on general reasoning tasks among open-source models and significantly outperforms open-source models in scientific domains, surpassing closed-source state-of-the-art models in professional tasks, such as molecular synthesis planning, reaction condition prediction, predicting thermodynamic stabilities for crystals. Our models are available at https://huggingface.co/internlm/Intern-S1.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1fd030f1c735042a", "title": "Position Bias Mitigates Position Bias:Mitigate Position Bias Through Inter-Position Knowledge Distillation", "authors": ["Yifei Wang", "Feng Xiong", "Yong Wang", "Linjing Li", "Xiangxiang Chu", "Daniel Dajun Zeng"], "year": 2025, "published": "2025-08-21T16:54:04Z", "updated": "2025-09-17T05:47:29Z", "arxiv_id": "http://arxiv.org/abs/2508.15709v2", "url": "https://arxiv.org/abs/2508.15709v2", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Long Context"], "benchmarks": [], "models": ["ying"], "key_results": ["By employing the Pos2Distill approach, we achieve enhanced uniformity and significant performance gains across all contextual positions in long-context retrieval and reasoning tasks.", "Crucially, both specialized systems exhibit strong cross-task generalization mutually, while achieving superior performance on their respective tasks."], "abstract": "Positional bias (PB), manifesting as non-uniform sensitivity across different contextual locations, significantly impairs long-context comprehension and processing capabilities. Previous studies have addressed PB either by modifying the underlying architectures or by employing extensive contextual awareness training. However, the former approach fails to effectively eliminate the substantial performance disparities, while the latter imposes significant data and computational overhead. To address PB effectively, we introduce \\textbf{Pos2Distill}, a position to position knowledge distillation framework. Pos2Distill transfers the superior capabilities from advantageous positions to less favorable ones, thereby reducing the huge performance gaps. The conceptual principle is to leverage the inherent, position-induced disparity to counteract the PB itself. We identify distinct manifestations of PB under \\textbf{\\textsc{r}}etrieval and \\textbf{\\textsc{r}}easoning paradigms, thereby designing two specialized instantiations: \\emph{Pos2Distill-R\\textsuperscript{1}} and \\emph{Pos2Distill-R\\textsuperscript{2}} respectively, both grounded in this core principle. By employing the Pos2Distill approach, we achieve enhanced uniformity and significant performance gains across all contextual positions in long-context retrieval and reasoning tasks. Crucially, both specialized systems exhibit strong cross-task generalization mutually, while achieving superior performance on their respective tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9c49b7a2cd4a6048", "title": "Overcoming the Communication-Performance Tradeoff in LLM Pretraining", "authors": ["Amir Sarfi", "Benjamin Thérien", "Joel Lidin", "Eugene Belilovsky"], "year": 2025, "published": "2025-08-21T16:48:19Z", "updated": "2026-07-23T09:18:31Z", "arxiv_id": "http://arxiv.org/abs/2508.15706v3", "url": "https://arxiv.org/abs/2508.15706v3", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "Quantization"], "benchmarks": [], "models": [], "key_results": ["While these local-update methods achieve communication reduction through reduced synchronization, they still require communicating dense model-sized pseudo-gradients, resulting in a communication bottleneck even for cross-datacenter links.", "In this work, we introduce SparseLoCo, a communication-efficient training algorithm for LLMs that can effectively leverage Top-k sparsification and 2-bit quantization to reach extreme sparsity in the communicated pseudo-gradient, as high as 97-99%, while achieving lower final loss than dense DiLoCo.", "In our empirical study of language model pre-training, we demonstrate that SparseLoCo's improvements over DiLoCo in performance and pseudo-gradient compression are maintained across dense model scales (178M-2B), an MoE transformer (645M-A273M), increasing the number of workers, and increasing communication intervals."], "abstract": "Communication-efficient distributed training algorithms (e.g., DiLoCo) have received considerable interest due to their benefits for training large language models (LLMs) in bandwidth-constrained settings, such as across datacenters and over the internet. While these local-update methods achieve communication reduction through reduced synchronization, they still require communicating dense model-sized pseudo-gradients, resulting in a communication bottleneck even for cross-datacenter links. While quantization is often applied to reduce the pseudo-gradient's size, in the context of LLM pre-training, existing approaches have not been able to leverage sparsification without incurring significant performance degradation. In this work, we introduce SparseLoCo, a communication-efficient training algorithm for LLMs that can effectively leverage Top-k sparsification and 2-bit quantization to reach extreme sparsity in the communicated pseudo-gradient, as high as 97-99%, while achieving lower final loss than dense DiLoCo. In our empirical study of language model pre-training, we demonstrate that SparseLoCo's improvements over DiLoCo in performance and pseudo-gradient compression are maintained across dense model scales (178M-2B), an MoE transformer (645M-A273M), increasing the number of workers, and increasing communication intervals.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ec78c462fabfaa8f", "title": "Beyond Imaging: Vision Transformer Digital Twin Surrogates for 3D+T Biological Tissue Dynamics", "authors": ["Kaan Berke Ugurlar", "Joaquín de Navascués", "Michael Taynnan Barros"], "year": 2025, "published": "2025-08-21T16:24:24Z", "updated": "2025-08-25T10:31:36Z", "arxiv_id": "http://arxiv.org/abs/2508.15883v2", "url": "https://arxiv.org/abs/2508.15883v2", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.AI", "cs.LG", "q-bio.TO"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Preference Optimization", "Interpretability"], "benchmarks": [], "models": ["phila", "ying"], "key_results": ["Evaluation across layers and biological replicates demonstrates VT-DTSN's robustness and consistency, achieving low error rates and high structural similarity while maintaining efficient inference through model optimization."], "abstract": "Understanding the dynamic organization and homeostasis of living tissues requires high-resolution, time-resolved imaging coupled with methods capable of extracting interpretable, predictive insights from complex datasets. Here, we present the Vision Transformer Digital Twin Surrogate Network (VT-DTSN), a deep learning framework for predictive modeling of 3D+T imaging data from biological tissue. By leveraging Vision Transformers pretrained with DINO (Self-Distillation with NO Labels) and employing a multi-view fusion strategy, VT-DTSN learns to reconstruct high-fidelity, time-resolved dynamics of a Drosophila midgut while preserving morphological and feature-level integrity across imaging depths. The model is trained with a composite loss prioritizing pixel-level accuracy, perceptual structure, and feature-space alignment, ensuring biologically meaningful outputs suitable for in silico experimentation and hypothesis testing. Evaluation across layers and biological replicates demonstrates VT-DTSN's robustness and consistency, achieving low error rates and high structural similarity while maintaining efficient inference through model optimization. This work establishes VT-DTSN as a feasible, high-fidelity surrogate for cross-timepoint reconstruction and for studying tissue dynamics, enabling computational exploration of cellular behaviors and homeostasis to complement time-resolved imaging studies in biological research.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6fe120629b8e5f01", "title": "TPLA: Tensor Parallel Latent Attention for Efficient Disaggregated Prefill and Decode Inference", "authors": ["Xiaojuan Tang", "Fanxu Meng", "Pingzhi Tang", "Yuxuan Wang", "Di Yin", "Xing Sun", "Muhan Zhang"], "year": 2025, "published": "2025-08-21T15:25:40Z", "updated": "2025-08-25T02:24:20Z", "arxiv_id": "http://arxiv.org/abs/2508.15881v2", "url": "https://arxiv.org/abs/2508.15881v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache", "Long Context", "Memory / RAG"], "benchmarks": ["DROP"], "models": ["DeepSeek-V2", "DeepSeek-V3", "V3", "ying", "yielding", "Kimi-K2"], "key_results": ["By reducing the per-device KV cache for DeepSeek-V3 and Kimi-K2, we achieve 1.79x and 1.93x speedups, respectively, at a 32K-token context length while maintaining performance on commonsense and LongBench benchmarks."], "abstract": "Multi-Head Latent Attention (MLA), introduced in DeepSeek-V2, compresses key-value states into a low-rank latent vector, caching only this vector to reduce memory. In tensor parallelism (TP), however, attention heads are computed across multiple devices, and each device must load the full cache, eroding the advantage of MLA over Grouped Query Attention (GQA). We propose Tensor-Parallel Latent Attention (TPLA): a scheme that partitions both the latent representation and each head's input dimension across devices, performs attention independently per shard, and then combines results with an all-reduce. TPLA preserves the benefits of a compressed KV cache while unlocking TP efficiency. Unlike Grouped Latent Attention (GLA), every head in TPLA still leverages the full latent representation, maintaining stronger representational capacity. TPLA is drop-in compatible with models pre-trained using MLA: it supports MLA-style prefilling and enables efficient tensor-parallel decoding without retraining. Applying simple orthogonal transforms -- e.g., the Hadamard transform or PCA -- before TP slicing further mitigates cross-shard interference, yielding minimal accuracy degradation. By reducing the per-device KV cache for DeepSeek-V3 and Kimi-K2, we achieve 1.79x and 1.93x speedups, respectively, at a 32K-token context length while maintaining performance on commonsense and LongBench benchmarks. TPLA can be implemented with FlashAttention-3, enabling practical end-to-end acceleration.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-280eb04e39d1d86d", "title": "ObjexMT: Objective Extraction and Metacognitive Calibration for LLM-as-a-Judge under Multi-Turn Jailbreaks", "authors": ["Hyunjun Kim", "Junwoo Ha", "Sangyoon Yu", "Haon Park"], "year": 2025, "published": "2025-08-23T03:32:04Z", "updated": "2025-10-08T21:13:00Z", "arxiv_id": "http://arxiv.org/abs/2508.16889v4", "url": "https://arxiv.org/abs/2508.16889v4", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Quantization", "Safety / Jailbreak"], "benchmarks": [], "models": ["gpt-4.1", "claude-sonnet-4", "claude-sonnet-4 offers the best selective risk and calibration", "gemini-2.5-flash", "Qwen3-235B-A22B-FP8", "deepseek-v3.1", "v3.1", "kimi-k2", "kimi-k2 achieves the highest objective-extraction accuracy"], "key_results": ["Across six models (gpt-4.1, claude-sonnet-4, Qwen3-235B-A22B-FP8, kimi-k2, deepseek-v3.1, gemini-2.5-flash) evaluated on SafeMTData\\_Attack600, SafeMTData\\_1K, and MHJ, kimi-k2 achieves the highest objective-extraction accuracy (0.612; 95\\% CI [0.594, 0.630]), while claude-sonnet-4 (0.603) and deepseek-v3.1 (0.599) are statistically tied."], "abstract": "LLM-as-a-Judge (LLMaaJ) enables scalable evaluation, yet we lack a decisive test of a judge's qualification: can it recover the hidden objective of a conversation and know when that inference is reliable? Large language models degrade with irrelevant or lengthy context, and multi-turn jailbreaks can scatter goals across turns. We present ObjexMT, a benchmark for objective extraction and metacognition. Given a multi-turn transcript, a model must output a one-sentence base objective and a self-reported confidence. Accuracy is scored by semantic similarity to gold objectives, then thresholded once on 300 calibration items ($τ^\\star = 0.66$; $F_1@τ^\\star = 0.891$). Metacognition is assessed with expected calibration error, Brier score, Wrong@High-Confidence (0.80 / 0.90 / 0.95), and risk--coverage curves. Across six models (gpt-4.1, claude-sonnet-4, Qwen3-235B-A22B-FP8, kimi-k2, deepseek-v3.1, gemini-2.5-flash) evaluated on SafeMTData\\_Attack600, SafeMTData\\_1K, and MHJ, kimi-k2 achieves the highest objective-extraction accuracy (0.612; 95\\% CI [0.594, 0.630]), while claude-sonnet-4 (0.603) and deepseek-v3.1 (0.599) are statistically tied. claude-sonnet-4 offers the best selective risk and calibration (AURC 0.242; ECE 0.206; Brier 0.254). Performance varies sharply across datasets (16--82\\% accuracy), showing that automated obfuscation imposes challenges beyond model choice. High-confidence errors remain: Wrong@0.90 ranges from 14.9\\% (claude-sonnet-4) to 47.7\\% (Qwen3-235B-A22B-FP8). ObjexMT therefore supplies an actionable test for LLM judges: when objectives are implicit, judges often misinfer them; exposing objectives or gating decisions by confidence is advisable. All experimental data are in the Supplementary Material and at https://github.com/hyunjun1121/ObjexMT_dataset.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-deb26ace65df20c2", "title": "UM3: Unsupervised Map to Map Matching", "authors": ["Chaolong Ying", "Yinan Zhang", "Lei Zhang", "Jiazhuang Wang", "Shujun Jia", "Tianshu Yu"], "year": 2025, "published": "2025-08-23T02:14:52Z", "updated": "2026-01-18T14:15:08Z", "arxiv_id": "http://arxiv.org/abs/2508.16874v2", "url": "https://arxiv.org/abs/2508.16874v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Experiments on real-world datasets demonstrate that our method achieves state-of-the-art accuracy in matching tasks, surpassing existing methods by a large margin, particularly in high-noise and large-scale scenarios."], "abstract": "Map-to-map matching is a critical task for aligning spatial data across heterogeneous sources, yet it remains challenging due to the lack of ground truth correspondences, sparse node features, and scalability demands. In this paper, we propose an unsupervised graph-based framework that addresses these challenges through three key innovations. First, our method is an unsupervised learning approach that requires no training data, which is crucial for large-scale map data where obtaining labeled training samples is challenging. Second, we introduce pseudo coordinates that capture the relative spatial layout of nodes within each map, which enhances feature discriminability and enables scale-invariant learning. Third, we design an mechanism to adaptively balance feature and geometric similarity, as well as a geometric-consistent loss function, ensuring robustness to noisy or incomplete coordinate data. At the implementation level, to handle large-scale maps, we develop a tile-based post-processing pipeline with overlapping regions and majority voting, which enables parallel processing while preserving boundary coherence. Experiments on real-world datasets demonstrate that our method achieves state-of-the-art accuracy in matching tasks, surpassing existing methods by a large margin, particularly in high-noise and large-scale scenarios. Our framework provides a scalable and practical solution for map alignment, offering a robust and efficient alternative to traditional approaches.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-dabd4ee0f1f1adee", "title": "Cyber Orbits of Large Scale Network Traffic", "authors": ["Jeremy Kepner", "Hayden Jananthan", "Chasen Milner", "Michael Houle", "Michael Jones", "Peter Michaleas", "Alex Pentland"], "year": 2025, "published": "2025-08-23T00:21:29Z", "updated": "2025-08-23T00:21:29Z", "arxiv_id": "http://arxiv.org/abs/2508.16847v1", "url": "https://arxiv.org/abs/2508.16847v1", "primary_category": "physics.soc-ph", "categories": ["physics.soc-ph", "cs.CR", "cs.NI"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["The advent of high-performance graph libraries, such as the GraphBLAS, has enabled the analysis of massive network data sets and revealed new models for their behavior."], "abstract": "The advent of high-performance graph libraries, such as the GraphBLAS, has enabled the analysis of massive network data sets and revealed new models for their behavior. Physical analogies for complicated network behavior can be a useful aid to understanding these newly discovered network phenomena. Prior work leveraged the canonical Gull's Lighthouse problem and developed a computational heuristic for modeling large scale network traffic using this model. A general solution using this approach requires overcoming the essential mathematical singularities in the resulting differential equations. Further investigation reveals a simpler physical interpretation that alleviates the need for solving challenging differential equations. Specifically, that the probability of observing a source at a temporal ``distance'' $r(t)$ at time $t$ is $p(t) \\propto 1/r(t)^2$. This analogy aligns with many physical phenomena and can be a rich source of intuition. Applying this physical analogy to the observed source correlations in the Anonymized Network Sensing Graph Challenge data leads to an elegant cyber orbit analogy that may assist with the understanding network behavior.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7a7b00eb13099dad", "title": "Cross-device Zero-shot Label Transfer via Alignment of Time Series Foundation Model Embeddings", "authors": ["Neal G. Ravindra", "Arijit Sehanobish"], "year": 2025, "published": "2025-08-22T23:22:41Z", "updated": "2025-08-22T23:22:41Z", "arxiv_id": "http://arxiv.org/abs/2509.06966v1", "url": "https://arxiv.org/abs/2509.06966v1", "primary_category": "eess.SP", "categories": ["eess.SP", "cs.AI", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["High-quality, medically validated labels exist for clinical actigraphy data but not for ubiquitous consumer wearables like the Apple Watch."], "abstract": "High-quality, medically validated labels exist for clinical actigraphy data but not for ubiquitous consumer wearables like the Apple Watch. Manually labeling wearables data is expensive and doesn't scale. This paper offers a novel framework that transfers valuable labels from a source domain (e.g., actigraphy) to a target domain (e.g., Apple Watch) without requiring paired data. Instead of working with raw time-series signals, we project both domains into a shared latent embedding space using time-series foundation models (TSFMs) and develop a new framework to align the cross-device representations. Our method, Adversarial Alignment of TSFM Embeddings forces the distributions of source and target embeddings to align within this space, facilitating label transfer across device type.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9aff163f7bf08cf3", "title": "Learning Explainable Imaging-Genetics Associations Related to a Neurological Disorder", "authors": ["Jueqi Wang", "Zachary Jacokes", "John Darrell Van Horn", "Michael C. Schatz", "Kevin A. Pelphrey", "Archana Venkataraman"], "year": 2025, "published": "2025-08-22T23:18:06Z", "updated": "2025-08-22T23:18:06Z", "arxiv_id": "http://arxiv.org/abs/2508.18303v1", "url": "https://arxiv.org/abs/2508.18303v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "q-bio.QM"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["Our results demonstrate that NeuroPathX outperforms competing baseline approaches and reveals biologically plausible associations linked to the disorder."], "abstract": "While imaging-genetics holds great promise for unraveling the complex interplay between brain structure and genetic variation in neurological disorders, traditional methods are limited to simplistic linear models or to black-box techniques that lack interpretability. In this paper, we present NeuroPathX, an explainable deep learning framework that uses an early fusion strategy powered by cross-attention mechanisms to capture meaningful interactions between structural variations in the brain derived from MRI and established biological pathways derived from genetics data. To enhance interpretability and robustness, we introduce two loss functions over the attention matrix - a sparsity loss that focuses on the most salient interactions and a pathway similarity loss that enforces consistent representations across the cohort. We validate NeuroPathX on both autism spectrum disorder and Alzheimer's disease. Our results demonstrate that NeuroPathX outperforms competing baseline approaches and reveals biologically plausible associations linked to the disorder. These findings underscore the potential of NeuroPathX to advance our understanding of complex brain disorders. Code is available at https://github.com/jueqiw/NeuroPathX .", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5ba0f6cc4593871e", "title": "Out of Distribution Detection for Efficient Continual Learning in Quality Prediction for Arc Welding", "authors": ["Yannik Hahn", "Jan Voets", "Antonin Koenigsfeld", "Hasan Tercan", "Tobias Meisen"], "year": 2025, "published": "2025-08-22T23:09:21Z", "updated": "2026-02-16T19:52:33Z", "arxiv_id": "http://arxiv.org/abs/2508.16832v2", "url": "https://arxiv.org/abs/2508.16832v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["In this work, we extend the VQ-VAE Transformer architecture - previously demonstrating state-of-the-art performance in weld quality prediction - by leveraging its autoregressive loss as a reliable out-of-distribution (OOD) detection mechanism."], "abstract": "Modern manufacturing relies heavily on fusion welding processes, including gas metal arc welding (GMAW). Despite significant advances in machine learning-based quality prediction, current models exhibit critical limitations when confronted with the inherent distribution shifts that occur in dynamic manufacturing environments. In this work, we extend the VQ-VAE Transformer architecture - previously demonstrating state-of-the-art performance in weld quality prediction - by leveraging its autoregressive loss as a reliable out-of-distribution (OOD) detection mechanism. Our approach exhibits superior performance compared to conventional reconstruction methods, embedding error-based techniques, and other established baselines. By integrating OOD detection with continual learning strategies, we optimize model adaptation, triggering updates only when necessary and thereby minimizing costly labeling requirements. We introduce a novel quantitative metric that simultaneously evaluates OOD detection capability while interpreting in-distribution performance. Experimental validation in real-world welding scenarios demonstrates that our framework effectively maintains robust quality prediction capabilities across significant distribution shifts, addressing critical challenges in dynamic manufacturing environments where process parameters frequently change. This research makes a substantial contribution to applied artificial intelligence by providing an explainable and at the same time adaptive solution for quality assurance in dynamic manufacturing processes - a crucial step towards robust, practical AI systems in the industrial environment.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e7e44f5b004be4d1", "title": "Latent Graph Learning in Generative Models of Neural Signals", "authors": ["Nathan X. Kodama", "Kenneth A. Loparo"], "year": 2025, "published": "2025-08-22T20:19:56Z", "updated": "2025-08-22T20:19:56Z", "arxiv_id": "http://arxiv.org/abs/2508.16776v1", "url": "https://arxiv.org/abs/2508.16776v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["By testing against numerical simulations of neural circuits with known ground-truth connectivity, we evaluate several hypotheses for explaining learned model weights."], "abstract": "Inferring temporal interaction graphs and higher-order structure from neural signals is a key problem in building generative models for systems neuroscience. Foundation models for large-scale neural data represent shared latent structures of neural signals. However, extracting interpretable latent graph representations in foundation models remains challenging and unsolved. Here we explore latent graph learning in generative models of neural signals. By testing against numerical simulations of neural circuits with known ground-truth connectivity, we evaluate several hypotheses for explaining learned model weights. We discover modest alignment between extracted network representations and the underlying directed graphs and strong alignment in the co-input graph representations. These findings motivate paths towards incorporating graph-based geometric constraints in the construction of large-scale foundation models for neural data.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d98a0506ca26bf9e", "title": "DR-CircuitGNN: Training Acceleration of Heterogeneous Circuit Graph Neural Network on GPUs", "authors": ["Yuebo Luo", "Shiyang Li", "Junran Tao", "Kiran Thorat", "Xi Xie", "Hongwu Peng", "Nuo Xu", "Caiwen Ding", "Shaoyi Huang"], "year": 2025, "published": "2025-08-22T20:05:38Z", "updated": "2025-08-22T20:05:38Z", "arxiv_id": "http://arxiv.org/abs/2508.16769v1", "url": "https://arxiv.org/abs/2508.16769v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["However, the improved representation capability comes at the cost of even higher computational complexity and processing cost due to their serial module-wise message-passing scheme, creating a significant performance bottleneck.", "Our experiments show that on three representative CircuitNet designs (small, medium, large), the proposed method can achieve up to 3.51x and 4.09x speedup compared to the SOTA for forward and backward propagation, respectively."], "abstract": "The increasing scale and complexity of integrated circuit design have led to increased challenges in Electronic Design Automation (EDA). Graph Neural Networks (GNNs) have emerged as a promising approach to assist EDA design as circuits can be naturally represented as graphs. While GNNs offer a foundation for circuit analysis, they often fail to capture the full complexity of EDA designs. Heterogeneous Graph Neural Networks (HGNNs) can better interpret EDA circuit graphs as they capture both topological relationships and geometric features. However, the improved representation capability comes at the cost of even higher computational complexity and processing cost due to their serial module-wise message-passing scheme, creating a significant performance bottleneck. In this paper, we propose DR-CircuitGNN, a fast GPU kernel design by leveraging row-wise sparsity-aware Dynamic-ReLU and optimizing SpMM kernels during heterogeneous message-passing to accelerate HGNNs training on EDA-related circuit graph datasets. To further enhance performance, we propose a parallel optimization strategy that maximizes CPU-GPU concurrency by concurrently processing independent subgraphs using multi-threaded CPU initialization and GPU kernel execution via multiple cudaStreams. Our experiments show that on three representative CircuitNet designs (small, medium, large), the proposed method can achieve up to 3.51x and 4.09x speedup compared to the SOTA for forward and backward propagation, respectively. On full-size CircuitNet and sampled Mini-CircuitNet, our parallel design enables up to 2.71x speed up over the official DGL implementation cuSPARSE with negligible impact on correlation scores and error rates.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-f11873fbb8524c7a", "title": "Toward Socially Aware Vision-Language Models: Evaluating Cultural Competence Through Multimodal Story Generation", "authors": ["Arka Mukherjee", "Shreya Ghosh"], "year": 2025, "published": "2025-08-22T19:39:02Z", "updated": "2025-08-22T19:39:02Z", "arxiv_id": "http://arxiv.org/abs/2508.16762v1", "url": "https://arxiv.org/abs/2508.16762v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.CY"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": ["phic"], "key_results": ["As Vision-Language Models (VLMs) achieve widespread deployment across diverse cultural contexts, ensuring their cultural competence becomes critical for responsible AI systems."], "abstract": "As Vision-Language Models (VLMs) achieve widespread deployment across diverse cultural contexts, ensuring their cultural competence becomes critical for responsible AI systems. While prior work has evaluated cultural awareness in text-only models and VLM object recognition tasks, no research has systematically assessed how VLMs adapt outputs when cultural identity cues are embedded in both textual prompts and visual inputs during generative tasks. We present the first comprehensive evaluation of VLM cultural competence through multimodal story generation, developing a novel multimodal framework that perturbs cultural identity and evaluates 5 contemporary VLMs on a downstream task: story generation. Our analysis reveals significant cultural adaptation capabilities, with rich culturally-specific vocabulary spanning names, familial terms, and geographic markers. However, we uncover concerning limitations: cultural competence varies dramatically across architectures, some models exhibit inverse cultural alignment, and automated metrics show architectural bias contradicting human assessments. Cross-modal evaluation shows that culturally distinct outputs are indeed detectable through visual-semantic similarity (28.7% within-nationality vs. 0.2% cross-nationality recall), yet visual-cultural understanding remains limited. In essence, we establish the promise and challenges of cultural competence in multimodal AI. We publicly release our codebase and data: https://github.com/ArkaMukherjee0/mmCultural", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1c71f89b5c6c18a7", "title": "Securing Heterogeneous Network (HetNet) Communications for Wildfire Management: Mitigating the Effects of Adversarial and Environmental Threats", "authors": ["Nesrine Benchoubane", "Olfa Ben Yahia", "William Ferguson", "Gurkan Gur", "Sumit Chakravarty", "Gregory Falco", "Gunes Karabulut Kurt"], "year": 2025, "published": "2025-08-22T19:38:31Z", "updated": "2025-08-22T19:38:31Z", "arxiv_id": "http://arxiv.org/abs/2508.16761v1", "url": "https://arxiv.org/abs/2508.16761v1", "primary_category": "cs.CR", "categories": ["cs.CR"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Moreover, increasing transmit power to counter environmental losses can inadvertently improve eavesdropper reception, thereby reducing overall link confidentiality."], "abstract": "In the face of adverse environmental conditions and cyber threats, robust communication systems for critical applications such as wildfire management and detection demand secure and resilient architectures. This paper presents a novel framework that considers both adversarial factors, building resilience into a heterogeneous network (HetNet) integrating Low Earth Orbit (LEO) satellite constellation with High-Altitude Platform Ground Stations (HAPGS) and Low-Altitude Platforms (LAPS), tailored to support wildfire management operations. Building upon our previous work on secure-by-component approach for link segment security, we extend protection to the communication layer by securing both Radio Frequency (RF)/Free Space Optics (FSO) management and different links. Through a case study, we quantify how environmental stressors impact secrecy capacity and expose the system to passive adversaries. Key findings demonstrate that atmospheric attenuation and beam misalignment can notably degrade secrecy capacity across both short- and long-range communication links, while high-altitude eavesdroppers face less signal degradation, increasing their interception capability. Moreover, increasing transmit power to counter environmental losses can inadvertently improve eavesdropper reception, thereby reducing overall link confidentiality. Our work not only highlights the importance of protecting networks from these dual threats but also aligns with the IEEE P3536 Standard for Space System Cybersecurity Design, ensuring resilience and the prevention of mission failures.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-4b1c72052f641c3f", "title": "GAICo: A Deployed and Extensible Framework for Evaluating Diverse and Multimodal Generative AI Outputs", "authors": ["Nitin Gupta", "Pallav Koppisetti", "Kausik Lakkaraju", "Biplav Srivastava"], "year": 2025, "published": "2025-08-22T19:13:21Z", "updated": "2026-01-22T19:46:27Z", "arxiv_id": "http://arxiv.org/abs/2508.16753v4", "url": "https://arxiv.org/abs/2508.16753v4", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["GAICo empowers AI researchers and developers to efficiently assess system performance, make evaluation reproducible, improve development velocity, and ultimately build more trustworthy AI systems, aligning with the goal of moving faster and safer in AI deployment."], "abstract": "The rapid proliferation of Generative AI (GenAI) into diverse, high-stakes domains necessitates robust and reproducible evaluation methods. However, practitioners often resort to ad-hoc, non-standardized scripts, as common metrics are often unsuitable for specialized, structured outputs (e.g., automated plans, time-series) or holistic comparison across modalities (e.g., text, audio, and image). This fragmentation hinders comparability and slows AI system development. To address this challenge, we present GAICo (Generative AI Comparator): a deployed, open-source Python library that streamlines and standardizes GenAI output comparison. GAICo provides a unified, extensible framework supporting a comprehensive suite of reference-based metrics for unstructured text, specialized structured data formats, and multimedia (images, audio). Its architecture features a high-level API for rapid, end-to-end analysis, from multi-model comparison to visualization and reporting, alongside direct metric access for granular control. We demonstrate GAICo's utility through a detailed case study evaluating and debugging complex, multi-modal AI Travel Assistant pipelines. GAICo empowers AI researchers and developers to efficiently assess system performance, make evaluation reproducible, improve development velocity, and ultimately build more trustworthy AI systems, aligning with the goal of moving faster and safer in AI deployment. Since its release on PyPI in Jun 2025, the tool has been downloaded over 16K times, across versions, by Dec 2025, demonstrating growing community interest.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d13f798e2957959c", "title": "Explainable AI for Predicting and Understanding Mathematics Achievement: A Cross-National Analysis of PISA 2018", "authors": ["Liu Liu", "Rui Dai"], "year": 2025, "published": "2025-08-22T19:02:15Z", "updated": "2025-08-22T19:02:15Z", "arxiv_id": "http://arxiv.org/abs/2508.16747v1", "url": "https://arxiv.org/abs/2508.16747v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CY", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": ["MATH"], "models": [], "key_results": ["This study applies explainable artificial intelligence (XAI) techniques to PISA 2018 data to predict math achievement and identify key predictors across ten countries (67,329 students).", "We tested four models: Multiple Linear Regression (MLR), Random Forest (RF), CATBoost, and Artificial Neural Networks (ANN), using student, family, and school variables.", "Non-linear models, especially RF and ANN, outperformed MLR, with RF balancing accuracy and generalizability."], "abstract": "Understanding the factors that shape students' mathematics performance is vital for designing effective educational policies. This study applies explainable artificial intelligence (XAI) techniques to PISA 2018 data to predict math achievement and identify key predictors across ten countries (67,329 students). We tested four models: Multiple Linear Regression (MLR), Random Forest (RF), CATBoost, and Artificial Neural Networks (ANN), using student, family, and school variables. Models were trained on 70% of the data (with 5-fold cross-validation) and tested on 30%, stratified by country. Performance was assessed with R^2 and Mean Absolute Error (MAE). To ensure interpretability, we used feature importance, SHAP values, and decision tree visualizations. Non-linear models, especially RF and ANN, outperformed MLR, with RF balancing accuracy and generalizability. Key predictors included socio-economic status, study time, teacher motivation, and students' attitudes toward mathematics, though their impact varied across countries. Visual diagnostics such as scatterplots of predicted vs actual scores showed RF and CATBoost aligned closely with actual performance. Findings highlight the non-linear and context-dependent nature of achievement and the value of XAI in educational research. This study uncovers cross-national patterns, informs equity-focused reforms, and supports the development of personalized learning strategies.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-de26995401b91b74", "title": "Learning the Language of Histopathology Images reveals Prognostic Subgroups in Invasive Lung Adenocarcinoma Patients", "authors": ["Abdul Rehman Akbar", "Usama Sajjad", "Ziyu Su", "Wencheng Li", "Fei Xing", "Jimmy Ruiz", "Wei Chen", "Muhammad Khalid Khan Niazi"], "year": 2025, "published": "2025-08-22T18:48:24Z", "updated": "2026-01-02T21:53:26Z", "arxiv_id": "http://arxiv.org/abs/2508.16742v2", "url": "https://arxiv.org/abs/2508.16742v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["phic", "yielded"], "key_results": ["In a multi-cohort dataset of 289 patients (600 slides), PathRosetta achieved an area under the curve (AUC) of 0.78 +- 0.04 on the internal cohort, significantly outperforming IASLC grading (AUC:0.71), AJCC staging (AUC:0.64), and other state-of-the-art AI models (AUC:0.62-0.67)."], "abstract": "Recurrence remains a major clinical challenge in surgically resected invasive lung adenocarcinoma, where existing grading and staging systems fail to capture the cellular complexity that underlies tumor aggressiveness. We present PathRosetta, a novel AI model that conceptualizes histopathology as a language, where cells serve as words, spatial neighborhoods form syntactic structures, and tissue architecture composes sentences. By learning this language of histopathology, PathRosetta predicts five-year recurrence directly from hematoxylin-and-eosin (H&E) slides, treating them as documents representing the state of the disease. In a multi-cohort dataset of 289 patients (600 slides), PathRosetta achieved an area under the curve (AUC) of 0.78 +- 0.04 on the internal cohort, significantly outperforming IASLC grading (AUC:0.71), AJCC staging (AUC:0.64), and other state-of-the-art AI models (AUC:0.62-0.67). It yielded a hazard ratio of 9.54 and a concordance index of 0.70, generalized robustly to external TCGA (AUC:0.75) and CPTAC (AUC:0.76) cohorts, and performed consistently across demographic and clinical subgroups. Beyond whole-slide prediction, PathRosetta uncovered prognostic subgroups within individual cell types, revealing that even within benign epithelial, stromal, or other cells, distinct morpho-spatial phenotypes correspond to divergent outcomes. Moreover, because the model explicitly understands what it is looking at, including cell types, cellular neighborhoods, and higher-order tissue morphology, it is inherently interpretable and can articulate the rationale behind its predictions. These findings establish that representing histopathology as a language enables interpretable and generalizable prognostication from routine histology.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-a4dac2c655998c69", "title": "Aligning Distributionally Robust Optimization with Practical Deep Learning Needs", "authors": ["Dmitrii Feoktistov", "Igor Ignashin", "Andrey Veprikov", "Nikita Borovko", "Alexander Bogdanov", "Savelii Chezhegov", "Aleksandr Beznosikov"], "year": 2025, "published": "2025-08-22T18:17:44Z", "updated": "2025-09-25T15:03:41Z", "arxiv_id": "http://arxiv.org/abs/2508.16734v2", "url": "https://arxiv.org/abs/2508.16734v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Empirical evaluation across diverse Deep Learning tasks, from Tabular DL to Split Learning tasks, demonstrates that ALSO outperforms both traditional optimizers and existing DRO methods."], "abstract": "While traditional Deep Learning (DL) optimization methods treat all training samples equally, Distributionally Robust Optimization (DRO) adaptively assigns importance weights to different samples. However, a significant gap exists between DRO and current DL practices. Modern DL optimizers require adaptivity and the ability to handle stochastic gradients, as these methods demonstrate superior performance. Additionally, for practical applications, a method should allow weight assignment not only to individual samples, but also to groups of objects (for example, all samples of the same class). This paper aims to bridge this gap by introducing ALSO $\\unicode{x2013}$ Adaptive Loss Scaling Optimizer $\\unicode{x2013}$ an adaptive algorithm for a modified DRO objective that can handle weight assignment to sample groups. We prove the convergence of our proposed algorithm for non-convex objectives, which is the typical case for DL models. Empirical evaluation across diverse Deep Learning tasks, from Tabular DL to Split Learning tasks, demonstrates that ALSO outperforms both traditional optimizers and existing DRO methods.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0c7e84b4bde0b4e8", "title": "Analysis of Transferability Estimation Metrics for Surgical Phase Recognition", "authors": ["Prabhant Singh", "Yiping Li", "Yasmina Al Khalil"], "year": 2025, "published": "2025-08-22T18:05:33Z", "updated": "2025-08-22T18:05:33Z", "arxiv_id": "http://arxiv.org/abs/2508.16730v1", "url": "https://arxiv.org/abs/2508.16730v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["ying", "yields"], "key_results": ["Fine-tuning pre-trained models has become a cornerstone of modern machine learning, allowing practitioners to achieve high performance with limited labeled data."], "abstract": "Fine-tuning pre-trained models has become a cornerstone of modern machine learning, allowing practitioners to achieve high performance with limited labeled data. In surgical video analysis, where expert annotations are especially time-consuming and costly, identifying the most suitable pre-trained model for a downstream task is both critical and challenging. Source-independent transferability estimation (SITE) offers a solution by predicting how well a model will fine-tune on target data using only its embeddings or outputs, without requiring full retraining. In this work, we formalize SITE for surgical phase recognition and provide the first comprehensive benchmark of three representative metrics, LogME, H-Score, and TransRate, on two diverse datasets (RAMIE and AutoLaparo). Our results show that LogME, particularly when aggregated by the minimum per-subset score, aligns most closely with fine-tuning accuracy; H-Score yields only weak predictive power; and TransRate often inverses true model rankings. Ablation studies show that when candidate models have similar performances, transferability estimates lose discriminative power, emphasizing the importance of maintaining model diversity or using additional validation. We conclude with practical guidelines for model selection and outline future directions toward domain-specific metrics, theoretical foundations, and interactive benchmarking tools.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9fe62f622fc29c09", "title": "Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse Autoencoders", "authors": ["David Chanin", "Adrià Garriga-Alonso"], "year": 2025, "published": "2025-08-22T17:26:33Z", "updated": "2026-07-07T00:22:17Z", "arxiv_id": "http://arxiv.org/abs/2508.16560v4", "url": "https://arxiv.org/abs/2508.16560v4", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["If L0 is too low, the SAE will mix correlated features to improve reconstruction."], "abstract": "Sparse Autoencoders (SAEs) extract features from LLM internal activations, meant to correspond to interpretable concepts. A core SAE training hyperparameter is L0: how many SAE features should fire per token on average. Existing work compares SAE algorithms using sparsity-reconstruction tradeoff plots, implying L0 is a free parameter with no inherently correct value aside from its effect on reconstruction. In this work we study the effect of L0 on SAEs, and show that if L0 is not set correctly, the SAE fails to disentangle the underlying features of the LLM. If L0 is too low, the SAE will mix correlated features to improve reconstruction. If L0 is too high, the SAE finds degenerate solutions that also mix features. Further, we present a proxy metric that can help guide the search for the correct L0 for an SAE on a given training distribution. We show that our method finds the correct L0 in toy models and coincides with peak sparse probing performance in LLM SAEs. We find that most commonly used SAEs have an L0 that is too low. Our work shows that practitioners must set L0 correctly to train SAEs with monosemantic features.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-772c8c1c39459292", "title": "HAMSA: Hijacking Aligned Compact Models via Stealthy Automation", "authors": ["Alexey Krylov", "Iskander Vagizov", "Dmitrii Korzh", "Maryam Douiba", "Azidine Guezzaz", "Vladimir Kokh", "Sergey D. Erokhin", "Elena V. Tutubalina", "Oleg Y. Rogov"], "year": 2025, "published": "2025-08-22T15:57:57Z", "updated": "2025-08-22T15:57:57Z", "arxiv_id": "http://arxiv.org/abs/2508.16484v1", "url": "https://arxiv.org/abs/2508.16484v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Large Language Models (LLMs), especially their compact efficiency-oriented variants, remain susceptible to jailbreak attacks that can elicit harmful outputs despite extensive alignment efforts."], "abstract": "Large Language Models (LLMs), especially their compact efficiency-oriented variants, remain susceptible to jailbreak attacks that can elicit harmful outputs despite extensive alignment efforts. Existing adversarial prompt generation techniques often rely on manual engineering or rudimentary obfuscation, producing low-quality or incoherent text that is easily flagged by perplexity-based filters. We present an automated red-teaming framework that evolves semantically meaningful and stealthy jailbreak prompts for aligned compact LLMs. The approach employs a multi-stage evolutionary search, where candidate prompts are iteratively refined using a population-based strategy augmented with temperature-controlled variability to balance exploration and coherence preservation. This enables the systematic discovery of prompts capable of bypassing alignment safeguards while maintaining natural language fluency. We evaluate our method on benchmarks in English (In-The-Wild Jailbreak Prompts on LLMs), and a newly curated Arabic one derived from In-The-Wild Jailbreak Prompts on LLMs and annotated by native Arabic linguists, enabling multilingual assessment.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-565e37e2e354eff5", "title": "LLM-as-classifier: Semi-Supervised, Iterative Framework for Hierarchical Text Classification using Large Language Models", "authors": ["Doohee You", "Andy Parisi", "Zach Vander Velden", "Lara Dantas Inojosa"], "year": 2025, "published": "2025-08-22T15:47:17Z", "updated": "2025-08-22T15:47:17Z", "arxiv_id": "http://arxiv.org/abs/2508.16478v1", "url": "https://arxiv.org/abs/2508.16478v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.IR"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["The advent of Large Language Models (LLMs) has provided unprecedented capabilities for analyzing unstructured text data."], "abstract": "The advent of Large Language Models (LLMs) has provided unprecedented capabilities for analyzing unstructured text data. However, deploying these models as reliable, robust, and scalable classifiers in production environments presents significant methodological challenges. Standard fine-tuning approaches can be resource-intensive and often struggle with the dynamic nature of real-world data distributions, which is common in the industry. In this paper, we propose a comprehensive, semi-supervised framework that leverages the zero- and few-shot capabilities of LLMs for building hierarchical text classifiers as a framework for a solution to these industry-wide challenges. Our methodology emphasizes an iterative, human-in-the-loop process that begins with domain knowledge elicitation and progresses through prompt refinement, hierarchical expansion, and multi-faceted validation. We introduce techniques for assessing and mitigating sequence-based biases and outline a protocol for continuous monitoring and adaptation. This framework is designed to bridge the gap between the raw power of LLMs and the practical need for accurate, interpretable, and maintainable classification systems in industry applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5cf7a4ca60257a95", "title": "A Probabilistic Inference Scaling Theory for LLM Self-Correction", "authors": ["Zhe Yang", "Yichang Zhang", "Yudong Wang", "Ziyao Xu", "Junyang Lin", "Zhifang Sui"], "year": 2025, "published": "2025-08-22T15:15:38Z", "updated": "2025-08-22T15:15:38Z", "arxiv_id": "http://arxiv.org/abs/2508.16456v1", "url": "https://arxiv.org/abs/2508.16456v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Large Language Models (LLMs) have demonstrated the capability to refine their generated answers through self-correction, enabling continuous performance improvement over multiple rounds.", "To fill this gap, we propose a probabilistic theory to model the dynamics of accuracy change and explain the performance improvements observed in multi-round self-correction."], "abstract": "Large Language Models (LLMs) have demonstrated the capability to refine their generated answers through self-correction, enabling continuous performance improvement over multiple rounds. However, the mechanisms underlying how and why accuracy evolves during this iterative process remain unexplored. To fill this gap, we propose a probabilistic theory to model the dynamics of accuracy change and explain the performance improvements observed in multi-round self-correction. Through mathematical derivation, we establish that the accuracy after the $t^{th}$ round of self-correction is given by: $Acc_t = Upp - α^t(Upp - Acc_0),$ where $Acc_0$ denotes the initial accuracy, $Upp$ represents the upper bound of accuracy convergence, and $α$ determines the rate of convergence. Based on our theory, these parameters can be calculated and the predicted accuracy curve then can be obtained through only a single round of self-correction. Extensive experiments across diverse models and datasets demonstrate that our theoretical predictions align closely with empirical accuracy curves, validating the effectiveness of the theory. Our work provides a theoretical foundation for understanding LLM self-correction, thus paving the way for further explorations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-82170ff5411e2ca4", "title": "Beyond Interpretability: Exploring the Comprehensibility of Adaptive Video Streaming through Large Language Models", "authors": ["Lianchen Jia", "Chaoyang Li", "Ziqi Yuan", "Jiahui Chen", "Tianchi Huang", "Jiangchuan Liu", "Lifeng Sun"], "year": 2025, "published": "2025-08-22T15:05:55Z", "updated": "2025-08-22T15:05:55Z", "arxiv_id": "http://arxiv.org/abs/2508.16448v1", "url": "https://arxiv.org/abs/2508.16448v1", "primary_category": "cs.MM", "categories": ["cs.MM", "cs.LG", "eess.IV"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["Experimental results demonstrate that \\texttt{ComTree} significantly improves comprehensibility while maintaining competitive performance, showing potential for further advancement."], "abstract": "Over the past decade, adaptive video streaming technology has witnessed significant advancements, particularly driven by the rapid evolution of deep learning techniques. However, the black-box nature of deep learning algorithms presents challenges for developers in understanding decision-making processes and optimizing for specific application scenarios. Although existing research has enhanced algorithm interpretability through decision tree conversion, interpretability does not directly equate to developers' subjective comprehensibility. To address this challenge, we introduce \\texttt{ComTree}, the first bitrate adaptation algorithm generation framework that considers comprehensibility. The framework initially generates the complete set of decision trees that meet performance requirements, then leverages large language models to evaluate these trees for developer comprehensibility, ultimately selecting solutions that best facilitate human understanding and enhancement. Experimental results demonstrate that \\texttt{ComTree} significantly improves comprehensibility while maintaining competitive performance, showing potential for further advancement. The source code is available at https://github.com/thu-media/ComTree.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0f67f39373a424a3", "title": "HunyuanVideo-Foley: Multimodal Diffusion with Representation Alignment for High-Fidelity Foley Audio Generation", "authors": ["Sizhe Shan", "Qiulin Li", "Yutao Cui", "Miles Yang", "Yuehai Wang", "Qun Yang", "Jin Zhou", "Zhao Zhong"], "year": 2025, "published": "2025-08-23T07:30:18Z", "updated": "2025-08-23T07:30:18Z", "arxiv_id": "http://arxiv.org/abs/2508.16930v1", "url": "https://arxiv.org/abs/2508.16930v1", "primary_category": "eess.AS", "categories": ["eess.AS", "cs.CV", "cs.SD"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Video Generation", "VLM"], "benchmarks": [], "models": ["HunyuanVideo-Foley", "hunyuanvideo-foley"], "key_results": ["Comprehensive evaluations demonstrate that HunyuanVideo-Foley achieves new state-of-the-art performance across audio fidelity, visual-semantic alignment, temporal alignment and distribution matching."], "abstract": "Recent advances in video generation produce visually realistic content, yet the absence of synchronized audio severely compromises immersion. To address key challenges in video-to-audio generation, including multimodal data scarcity, modality imbalance and limited audio quality in existing methods, we propose HunyuanVideo-Foley, an end-to-end text-video-to-audio framework that synthesizes high-fidelity audio precisely aligned with visual dynamics and semantic context. Our approach incorporates three core innovations: (1) a scalable data pipeline curating 100k-hour multimodal datasets through automated annotation; (2) a representation alignment strategy using self-supervised audio features to guide latent diffusion training, efficiently improving audio quality and generation stability; (3) a novel multimodal diffusion transformer resolving modal competition, containing dual-stream audio-video fusion through joint attention, and textual semantic injection via cross-attention. Comprehensive evaluations demonstrate that HunyuanVideo-Foley achieves new state-of-the-art performance across audio fidelity, visual-semantic alignment, temporal alignment and distribution matching. The demo page is available at: https://szczesnys.github.io/hunyuanvideo-foley/.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b6241a1afd30f10a", "title": "MDD: A Dataset for Text-and-Music Conditioned Duet Dance Generation", "authors": ["Prerit Gupta", "Jason Alexander Fotso-Puepi", "Zhengyuan Li", "Jay Mehta", "Aniket Bera"], "year": 2025, "published": "2025-08-23T05:56:37Z", "updated": "2025-08-23T05:56:37Z", "arxiv_id": "http://arxiv.org/abs/2508.16911v1", "url": "https://arxiv.org/abs/2508.16911v1", "primary_category": "cs.GR", "categories": ["cs.GR", "cs.CV", "cs.MM", "cs.SD"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["We introduce Multimodal DuetDance (MDD), a diverse multimodal benchmark dataset designed for text-controlled and music-conditioned 3D duet dance motion generation."], "abstract": "We introduce Multimodal DuetDance (MDD), a diverse multimodal benchmark dataset designed for text-controlled and music-conditioned 3D duet dance motion generation. Our dataset comprises 620 minutes of high-quality motion capture data performed by professional dancers, synchronized with music, and detailed with over 10K fine-grained natural language descriptions. The annotations capture a rich movement vocabulary, detailing spatial relationships, body movements, and rhythm, making MDD the first dataset to seamlessly integrate human motions, music, and text for duet dance generation. We introduce two novel tasks supported by our dataset: (1) Text-to-Duet, where given music and a textual prompt, both the leader and follower dance motion are generated (2) Text-to-Dance Accompaniment, where given music, textual prompt, and the leader's motion, the follower's motion is generated in a cohesive, text-aligned manner. We include baseline evaluations on both tasks to support future research.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-64df30f230fb1f8b", "title": "AMMKD: Adaptive Multimodal Multi-teacher Distillation for Lightweight Vision-Language Models", "authors": ["Yuqi Li", "Chuanguang Yang", "Junhao Dong", "Zhengtao Yao", "Haoyan Xu", "Zeyu Dong", "Hansheng Zeng", "Zhulin An", "Yingli Tian"], "year": 2025, "published": "2025-08-23T04:52:20Z", "updated": "2025-08-23T04:52:20Z", "arxiv_id": "http://arxiv.org/abs/2509.00039v1", "url": "https://arxiv.org/abs/2509.00039v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Distillation", "VLM"], "benchmarks": [], "models": [], "key_results": ["To reduce model parameters and further improve performance, we design a multi-teacher knowledge distillation framework to pre-train two CLIP teacher models.", "Extensive experiments on three benchmark datasets demonstrate that AMMKD achieves superior performance while significantly reducing model complexity, validating its effectiveness and flexibility."], "abstract": "The success of large-scale visual language pretraining (VLP) models has driven widespread adoption of image-text retrieval tasks. However, their deployment on mobile devices remains limited due to large model sizes and computational complexity. We propose Adaptive Multi-Modal Multi-Teacher Knowledge Distillation (AMMKD), a novel framework that integrates multi-modal feature fusion, multi-teacher distillation, and adaptive optimization to deliver lightweight yet effective retrieval models. Specifically, our method begins with a feature fusion network that extracts and merges discriminative features from both the image and text modalities. To reduce model parameters and further improve performance, we design a multi-teacher knowledge distillation framework to pre-train two CLIP teacher models. We decouple modalities by pre-computing and storing text features as class vectors via the teacher text encoder to enhance efficiency. To better align teacher and student outputs, we apply KL scatter for probability distribution matching. Finally, we design an adaptive dynamic weighting scheme that treats multi-teacher distillation as a multi-objective optimization problem. By leveraging gradient space diversity, we dynamically adjust the influence of each teacher, reducing conflicts and guiding the student toward more optimal learning directions. Extensive experiments on three benchmark datasets demonstrate that AMMKD achieves superior performance while significantly reducing model complexity, validating its effectiveness and flexibility.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9b35f6cdf57dad84", "title": "Multimodal Medical Endoscopic Image Analysis via Progressive Disentangle-aware Contrastive Learning", "authors": ["Junhao Wu", "Yun Li", "Junhao Li", "Jingliang Bian", "Xiaomao Fan", "Wenbin Lei", "Ruxin Wang"], "year": 2025, "published": "2025-08-23T03:02:51Z", "updated": "2025-08-23T03:02:51Z", "arxiv_id": "http://arxiv.org/abs/2508.16882v1", "url": "https://arxiv.org/abs/2508.16882v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["Comprehensive experiments on multiple datasets demonstrate that our method consistently outperforms state-of-the-art approaches, achieving superior accuracy across diverse real clinical scenarios."], "abstract": "Accurate segmentation of laryngo-pharyngeal tumors is crucial for precise diagnosis and effective treatment planning. However, traditional single-modality imaging methods often fall short of capturing the complex anatomical and pathological features of these tumors. In this study, we present an innovative multi-modality representation learning framework based on the `Align-Disentangle-Fusion' mechanism that seamlessly integrates 2D White Light Imaging (WLI) and Narrow Band Imaging (NBI) pairs to enhance segmentation performance. A cornerstone of our approach is multi-scale distribution alignment, which mitigates modality discrepancies by aligning features across multiple transformer layers. Furthermore, a progressive feature disentanglement strategy is developed with the designed preliminary disentanglement and disentangle-aware contrastive learning to effectively separate modality-specific and shared features, enabling robust multimodal contrastive learning and efficient semantic fusion. Comprehensive experiments on multiple datasets demonstrate that our method consistently outperforms state-of-the-art approaches, achieving superior accuracy across diverse real clinical scenarios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-40dd3b00adbef218", "title": "Multimodal LLMs See Sentiment", "authors": ["Neemias B. da Silva", "John Harrison", "Rodrigo Minetto", "Myriam R. Delgado", "Bogdan T. Nassu", "Thiago H. Silva"], "year": 2025, "published": "2025-08-23T02:11:46Z", "updated": "2026-05-27T19:57:46Z", "arxiv_id": "http://arxiv.org/abs/2508.16873v3", "url": "https://arxiv.org/abs/2508.16873v3", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.SI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Experiments on a recent benchmark show that a two-stage MLLM description-mediated pipeline can substantially improve prediction accuracy under several evaluation settings, particularly when the LLM component is fine-tuned.", "Across different agreement thresholds and sentiment granularities, the strongest configurations of this pipeline outperform lexicon-, CNN-, and Transformer-based baselines in our benchmark by up to 30.9%, 64.8%, and 42.4%, respectively.", "In cross-dataset evaluation, the proposed pipeline - without training or fine-tuning on the target dataset - still surpasses the best in-domain baseline by over 8%."], "abstract": "Understanding how visual content conveys sentiment is increasingly important in a digital landscape dominated by imagery. However, sentiment perception depends on complex scene-level semantics, making this a challenging task for computational models. This paper examines how Multimodal Large Language Models (MLLMs) perform sentiment analysis in images through a systematic, evaluation-driven study encompassing three perspectives: (i) direct sentiment classification from images using MLLMs; (ii) sentiment analysis on MLLM-generated descriptions using pre-trained LLMs; and (iii) fine-tuning these LLMs on sentiment-labeled descriptions to assess performance and generalization. Experiments on a recent benchmark show that a two-stage MLLM description-mediated pipeline can substantially improve prediction accuracy under several evaluation settings, particularly when the LLM component is fine-tuned. Across different agreement thresholds and sentiment granularities, the strongest configurations of this pipeline outperform lexicon-, CNN-, and Transformer-based baselines in our benchmark by up to 30.9%, 64.8%, and 42.4%, respectively. In cross-dataset evaluation, the proposed pipeline - without training or fine-tuning on the target dataset - still surpasses the best in-domain baseline by over 8%. Overall, the study provides a comprehensive assessment of MLLM description-mediated sentiment analysis, clarifying the conditions under which it is effective, the scenarios in which it fails, and its comparison with traditional vision-based approaches, while also providing a reproducible benchmark resource for future research.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0e89c681553bd7d7", "title": "Beyond Emotion Recognition: A Multi-Turn Multimodal Emotion Understanding and Reasoning Benchmark", "authors": ["Jinpeng Hu", "Hongchang Shi", "Chongyuan Dai", "Zhuo Li", "Peipei Song", "Meng Wang"], "year": 2025, "published": "2025-08-23T01:10:29Z", "updated": "2025-08-23T01:10:29Z", "arxiv_id": "http://arxiv.org/abs/2508.16859v1", "url": "https://arxiv.org/abs/2508.16859v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Multi-Agent", "VLM"], "benchmarks": [], "models": [], "key_results": ["Besides, we propose a multi-agent framework, where each agent specializes in a specific aspect, such as background context, character dynamics, and event details, to improve the system's reasoning capabilities."], "abstract": "Multimodal large language models (MLLMs) have been widely applied across various fields due to their powerful perceptual and reasoning capabilities. In the realm of psychology, these models hold promise for a deeper understanding of human emotions and behaviors. However, recent research primarily focuses on enhancing their emotion recognition abilities, leaving the substantial potential in emotion reasoning, which is crucial for improving the naturalness and effectiveness of human-machine interactions. Therefore, in this paper, we introduce a multi-turn multimodal emotion understanding and reasoning (MTMEUR) benchmark, which encompasses 1,451 video data from real-life scenarios, along with 5,101 progressive questions. These questions cover various aspects, including emotion recognition, potential causes of emotions, future action prediction, etc. Besides, we propose a multi-agent framework, where each agent specializes in a specific aspect, such as background context, character dynamics, and event details, to improve the system's reasoning capabilities. Furthermore, we conduct experiments with existing MLLMs and our agent-based method on the proposed benchmark, revealing that most models face significant challenges with this task.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9d2342f3d7fe38cd", "title": "Towards Open-Vocabulary Multimodal 3D Object Detection with Attributes", "authors": ["Xinhao Xiang", "Kuan-Chuan Peng", "Suhas Lohit", "Michael J. Jones", "Jiawei Zhang"], "year": 2025, "published": "2025-08-22T22:02:49Z", "updated": "2025-08-22T22:02:49Z", "arxiv_id": "http://arxiv.org/abs/2508.16812v1", "url": "https://arxiv.org/abs/2508.16812v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Our results on both the nuScenes and Argoverse 2 datasets show that under the condition of no given anchor sizes of novel classes, OVODA outperforms the state-of-the-art methods in open-vocabulary 3D object detection while successfully recognizing object attributes."], "abstract": "3D object detection plays a crucial role in autonomous systems, yet existing methods are limited by closed-set assumptions and struggle to recognize novel objects and their attributes in real-world scenarios. We propose OVODA, a novel framework enabling both open-vocabulary 3D object and attribute detection with no need to know the novel class anchor size. OVODA uses foundation models to bridge the semantic gap between 3D features and texts while jointly detecting attributes, e.g., spatial relationships, motion states, etc. To facilitate such research direction, we propose OVAD, a new dataset that supplements existing 3D object detection benchmarks with comprehensive attribute annotations. OVODA incorporates several key innovations, including foundation model feature concatenation, prompt tuning strategies, and specialized techniques for attribute detection, including perspective-specified prompts and horizontal flip augmentation. Our results on both the nuScenes and Argoverse 2 datasets show that under the condition of no given anchor sizes of novel classes, OVODA outperforms the state-of-the-art methods in open-vocabulary 3D object detection while successfully recognizing object attributes. Our OVAD dataset is released here: https://doi.org/10.5281/zenodo.16904069 .", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1df56eda47473b50", "title": "Leveraging Language Models and Machine Learning in Verbal Autopsy Analysis", "authors": ["Yue Chu"], "year": 2025, "published": "2025-08-22T21:40:54Z", "updated": "2025-08-22T21:40:54Z", "arxiv_id": "http://arxiv.org/abs/2508.19274v1", "url": "https://arxiv.org/abs/2508.19274v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Using empirical data from South Africa, we demonstrate that with the narrative alone, transformer-based PLMs with task-specific fine-tuning outperform leading question-only algorithms at both the individual and population levels, particularly in identifying non-communicable diseases.", "Multimodal approaches further improve performance in COD classification, confirming that each modality has unique contributions and may capture valuable information that is not present in the other modality."], "abstract": "In countries without civil registration and vital statistics, verbal autopsy (VA) is a critical tool for estimating cause of death (COD) and inform policy priorities. In VA, interviewers ask proximal informants for details on the circumstances preceding a death, in the form of unstructured narratives and structured questions. Existing automated VA cause classification algorithms only use the questions and ignore the information in the narratives. In this thesis, we investigate how the VA narrative can be used for automated COD classification using pretrained language models (PLMs) and machine learning (ML) techniques. Using empirical data from South Africa, we demonstrate that with the narrative alone, transformer-based PLMs with task-specific fine-tuning outperform leading question-only algorithms at both the individual and population levels, particularly in identifying non-communicable diseases. We explore various multimodal fusion strategies combining narratives and questions in unified frameworks. Multimodal approaches further improve performance in COD classification, confirming that each modality has unique contributions and may capture valuable information that is not present in the other modality. We also characterize physician-perceived information sufficiency in VA. We describe variations in sufficiency levels by age and COD and demonstrate that classification accuracy is affected by sufficiency for both physicians and models. Overall, this thesis advances the growing body of knowledge at the intersection of natural language processing, epidemiology, and global health. It demonstrates the value of narrative in enhancing COD classification. Our findings underscore the need for more high-quality data from more diverse settings to use in training and fine-tuning PLM/ML methods, and offer valuable insights to guide the rethinking and redesign of the VA instrument and interview.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-555fcd51ca6ea62f", "title": "WebMMU: A Benchmark for Multimodal Multilingual Website Understanding and Code Generation", "authors": ["Rabiul Awal", "Mahsa Massoud", "Aarash Feizi", "Zichao Li", "Suyuchen Wang", "Christopher Pal", "Aishwarya Agrawal", "David Vazquez", "Siva Reddy", "Juan A. Rodriguez", "Perouz Taslakian", "Spandana Gella", "Sai Rajeswar"], "year": 2025, "published": "2025-08-22T19:41:02Z", "updated": "2025-08-22T19:41:02Z", "arxiv_id": "http://arxiv.org/abs/2508.16763v1", "url": "https://arxiv.org/abs/2508.16763v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["These findings reveal key limitations in current MLLMs and underscore the need for improved multimodal and cross-lingual reasoning to build future web agents capable of automating diverse web development tasks."], "abstract": "We present WebMMU, a multilingual benchmark that evaluates three core web tasks: (1) website visual question answering, (2) code editing involving HTML/CSS/JavaScript, and (3) mockup-to-code generation. Unlike prior benchmarks that treat these tasks separately, WebMMU unifies them using expert-annotated, real-world web data to assess models' abilities in complex multi-step reasoning, precise element grounding, and functional UI comprehension and coding. Our evaluation shows that while multimodal large language models (MLLMs) perform well on basic information extraction, they struggle with reasoning and grounding, editing code to preserve functionality, and generating design-to-code that maintains hierarchy and supports multilingual content. These findings reveal key limitations in current MLLMs and underscore the need for improved multimodal and cross-lingual reasoning to build future web agents capable of automating diverse web development tasks.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-92c0ec3c46c93233", "title": "A Framework for Benchmarking Fairness-Utility Trade-offs in Text-to-Image Models via Pareto Frontiers", "authors": ["Marco N. Bochernitsan", "Rodrigo C. Barros", "Lucas S. Kupssinskü"], "year": 2025, "published": "2025-08-22T19:09:22Z", "updated": "2025-08-22T19:09:22Z", "arxiv_id": "http://arxiv.org/abs/2508.16752v1", "url": "https://arxiv.org/abs/2508.16752v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": ["Stable Diffusion", "FLUX text-to-image models. Our method shows that most default hyperparameterizations of the text-to-image model are dominated solutions in the fairness-utility space"], "key_results": ["Achieving fairness in text-to-image generation demands mitigating social biases without compromising visual fidelity, a challenge critical to responsible AI."], "abstract": "Achieving fairness in text-to-image generation demands mitigating social biases without compromising visual fidelity, a challenge critical to responsible AI. Current fairness evaluation procedures for text-to-image models rely on qualitative judgment or narrow comparisons, which limit the capacity to assess both fairness and utility in these models and prevent reproducible assessment of debiasing methods. Existing approaches typically employ ad-hoc, human-centered visual inspections that are both error-prone and difficult to replicate. We propose a method for evaluating fairness and utility in text-to-image models using Pareto-optimal frontiers across hyperparametrization of debiasing methods. Our method allows for comparison between distinct text-to-image models, outlining all configurations that optimize fairness for a given utility and vice-versa. To illustrate our evaluation method, we use Normalized Shannon Entropy and ClipScore for fairness and utility evaluation, respectively. We assess fairness and utility in Stable Diffusion, Fair Diffusion, SDXL, DeCoDi, and FLUX text-to-image models. Our method shows that most default hyperparameterizations of the text-to-image model are dominated solutions in the fairness-utility space, and it is straightforward to find better hyperparameters.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9b275d15d497eb10", "title": "FAIRWELL: Fair Multimodal Self-Supervised Learning for Wellbeing Prediction", "authors": ["Jiaee Cheong", "Abtin Mogharabin", "Paul Liang", "Hatice Gunes", "Sinan Kalkan"], "year": 2025, "published": "2025-08-22T19:03:06Z", "updated": "2025-08-22T19:03:06Z", "arxiv_id": "http://arxiv.org/abs/2508.16748v1", "url": "https://arxiv.org/abs/2508.16748v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Early efforts on leveraging self-supervised learning (SSL) to improve machine learning (ML) fairness has proven promising.", "Leveraging on this, we propose a novel subject-level loss function to learn fairer representations via the following three mechanisms, adapting the variance-invariance-covariance regularization (VICReg) method: (i) the variance term, which reduces reliance on the protected attribute as a trivial solution; (ii) the invariance term, which ensures consistent predictions for similar individuals; and (iii) the covariance term, which minimizes correlational dependence on the protected attribute.", "Our findings indicate that our framework improves overall fairness performance with minimal reduction in classification performance and significantly improves on the performance-fairness Pareto frontier."], "abstract": "Early efforts on leveraging self-supervised learning (SSL) to improve machine learning (ML) fairness has proven promising. However, such an approach has yet to be explored within a multimodal context. Prior work has shown that, within a multimodal setting, different modalities contain modality-unique information that can complement information of other modalities. Leveraging on this, we propose a novel subject-level loss function to learn fairer representations via the following three mechanisms, adapting the variance-invariance-covariance regularization (VICReg) method: (i) the variance term, which reduces reliance on the protected attribute as a trivial solution; (ii) the invariance term, which ensures consistent predictions for similar individuals; and (iii) the covariance term, which minimizes correlational dependence on the protected attribute. Consequently, our loss function, coined as FAIRWELL, aims to obtain subject-independent representations, enforcing fairness in multimodal prediction tasks. We evaluate our method on three challenging real-world heterogeneous healthcare datasets (i.e. D-Vlog, MIMIC and MODMA) which contain different modalities of varying length and different prediction tasks. Our findings indicate that our framework improves overall fairness performance with minimal reduction in classification performance and significantly improves on the performance-fairness Pareto frontier.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e5b56065cdf4c4b3", "title": "Hyperbolic Multimodal Representation Learning for Biological Taxonomies", "authors": ["ZeMing Gong", "Chuanqi Tang", "Xiaoliang Huo", "Nicholas Pellegrino", "Austin T. Wang", "Graham W. Taylor", "Angel X. Chang", "Scott C. Lowe", "Joakim Bruslund Haurum"], "year": 2025, "published": "2025-08-22T18:52:50Z", "updated": "2025-08-22T18:52:50Z", "arxiv_id": "http://arxiv.org/abs/2508.16744v1", "url": "https://arxiv.org/abs/2508.16744v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Experiments on the BIOSCAN-1M dataset show that hyperbolic embedding achieves competitive performance with Euclidean baselines, and outperforms all other models on unseen species classification using DNA barcodes."], "abstract": "Taxonomic classification in biodiversity research involves organizing biological specimens into structured hierarchies based on evidence, which can come from multiple modalities such as images and genetic information. We investigate whether hyperbolic networks can provide a better embedding space for such hierarchical models. Our method embeds multimodal inputs into a shared hyperbolic space using contrastive and a novel stacked entailment-based objective. Experiments on the BIOSCAN-1M dataset show that hyperbolic embedding achieves competitive performance with Euclidean baselines, and outperforms all other models on unseen species classification using DNA barcodes. However, fine-grained classification and open-world generalization remain challenging. Our framework offers a structure-aware foundation for biodiversity modelling, with potential applications to species discovery, ecological monitoring, and conservation efforts.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-548680814c319b7b", "title": "A Disease-Centric Vision-Language Foundation Model for Precision Oncology in Kidney Cancer", "authors": ["Yuhui Tao", "Zhongwei Zhao", "Zilong Wang", "Xufang Luo", "Feng Chen", "Kang Wang", "Chuanfu Wu", "Xue Zhang", "Shaoting Zhang", "Jiaxi Yao", "Xingwei Jin", "Xinyang Jiang", "Yifan Yang", "Dongsheng Li", "Lili Qiu", "Zhiqiang Shao", "Jianming Guo", "Nengwang Yu", "Shuo Wang", "Ying Xiong"], "year": 2025, "published": "2025-08-22T17:48:19Z", "updated": "2025-08-22T17:48:19Z", "arxiv_id": "http://arxiv.org/abs/2508.16569v1", "url": "https://arxiv.org/abs/2508.16569v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.AI", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["RenalCLIP achieved better performance and superior generalizability across 10 core tasks spanning the full clinical workflow of kidney cancer, including anatomical assessment, diagnostic classification, and survival prediction, compared with other state-of-the-art general-purpose CT foundation models.", "Especially, for complicated task like recurrence-free survival prediction in the TCIA cohort, RenalCLIP achieved a C-index of 0.726, representing a substantial improvement of approximately 20% over the leading baselines.", "Furthermore, RenalCLIP's pre-training imparted remarkable data efficiency; in the diagnostic classification task, it only needs 20% training data to achieve the peak performance of all baseline models even after they were fully fine-tuned on 100% of the data."], "abstract": "The non-invasive assessment of increasingly incidentally discovered renal masses is a critical challenge in urologic oncology, where diagnostic uncertainty frequently leads to the overtreatment of benign or indolent tumors. In this study, we developed and validated RenalCLIP using a dataset of 27,866 CT scans from 8,809 patients across nine Chinese medical centers and the public TCIA cohort, a visual-language foundation model for characterization, diagnosis and prognosis of renal mass. The model was developed via a two-stage pre-training strategy that first enhances the image and text encoders with domain-specific knowledge before aligning them through a contrastive learning objective, to create robust representations for superior generalization and diagnostic precision. RenalCLIP achieved better performance and superior generalizability across 10 core tasks spanning the full clinical workflow of kidney cancer, including anatomical assessment, diagnostic classification, and survival prediction, compared with other state-of-the-art general-purpose CT foundation models. Especially, for complicated task like recurrence-free survival prediction in the TCIA cohort, RenalCLIP achieved a C-index of 0.726, representing a substantial improvement of approximately 20% over the leading baselines. Furthermore, RenalCLIP's pre-training imparted remarkable data efficiency; in the diagnostic classification task, it only needs 20% training data to achieve the peak performance of all baseline models even after they were fully fine-tuned on 100% of the data. Additionally, it achieved superior performance in report generation, image-text retrieval and zero-shot diagnosis tasks. Our findings establish that RenalCLIP provides a robust tool with the potential to enhance diagnostic accuracy, refine prognostic stratification, and personalize the management of patients with kidney cancer.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-8b5759fbff106cee", "title": "Can VLMs Recall Factual Associations From Visual References?", "authors": ["Dhananjay Ashok", "Ashutosh Chaubey", "Hirona J. Arai", "Jonathan May", "Jesse Thomason"], "year": 2025, "published": "2025-08-22T16:47:37Z", "updated": "2025-08-22T16:47:37Z", "arxiv_id": "http://arxiv.org/abs/2508.18297v1", "url": "https://arxiv.org/abs/2508.18297v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.CL"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["We show that such linking failures are correlated with the expression of distinct patterns in model internal states, and that probes on these internal states achieve over 92% accuracy at flagging cases where the VLM response is unreliable."], "abstract": "Through a controlled study, we identify a systematic deficiency in the multimodal grounding of Vision Language Models (VLMs). While VLMs can recall factual associations when provided a textual reference to an entity; their ability to do so is significantly diminished when the reference is visual instead. Forcing VLMs to rely on image representations of an entity halves their ability to recall factual knowledge, suggesting that VLMs struggle to link their internal knowledge of an entity with its image representation. We show that such linking failures are correlated with the expression of distinct patterns in model internal states, and that probes on these internal states achieve over 92% accuracy at flagging cases where the VLM response is unreliable. These probes can be applied, without retraining, to identify when a VLM will fail to correctly answer a question that requires an understanding of multimodal input. When used to facilitate selective prediction on a visual question answering task, the probes increase coverage by 7.87% (absolute) while also reducing the risk of error by 0.9% (absolute). Addressing the systematic, detectable deficiency is an important avenue in language grounding, and we provide informed recommendations for future directions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-68d8a7102264b38b", "title": "Seeing Clearly, Forgetting Deeply: Revisiting Fine-Tuned Video Generators for Driving Simulation", "authors": ["Chun-Peng Chang", "Chen-Yu Wang", "Julian Schmidt", "Holger Caesar", "Alain Pagani"], "year": 2025, "published": "2025-08-22T16:35:19Z", "updated": "2025-08-22T16:35:19Z", "arxiv_id": "http://arxiv.org/abs/2508.16512v1", "url": "https://arxiv.org/abs/2508.16512v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "World Model", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Recent advancements in video generation have substantially improved visual quality and temporal coherence, making these models increasingly appealing for applications such as autonomous driving, particularly in the context of driving simulation and so-called \"world models\".", "In this work, we investigate the effects of existing fine-tuning video generation approaches on structured driving datasets and uncover a potential trade-off: although visual fidelity improves, spatial accuracy in modeling dynamic elements may degrade.", "However, the very regular and repetitive nature of driving scenes allows visual quality to improve by modeling dominant scene motion patterns, without necessarily preserving fine-grained dynamic behavior."], "abstract": "Recent advancements in video generation have substantially improved visual quality and temporal coherence, making these models increasingly appealing for applications such as autonomous driving, particularly in the context of driving simulation and so-called \"world models\". In this work, we investigate the effects of existing fine-tuning video generation approaches on structured driving datasets and uncover a potential trade-off: although visual fidelity improves, spatial accuracy in modeling dynamic elements may degrade. We attribute this degradation to a shift in the alignment between visual quality and dynamic understanding objectives. In datasets with diverse scene structures within temporal space, where objects or perspective shift in varied ways, these objectives tend to highly correlated. However, the very regular and repetitive nature of driving scenes allows visual quality to improve by modeling dominant scene motion patterns, without necessarily preserving fine-grained dynamic behavior. As a result, fine-tuning encourages the model to prioritize surface-level realism over dynamic accuracy. To further examine this phenomenon, we show that simple continual learning strategies, such as replay from diverse domains, can offer a balanced alternative by preserving spatial accuracy while maintaining strong visual quality.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9e7fd1b7f284b658", "title": "Sentiment-Aware Mean-Variance Portfolio Optimization for Cryptocurrencies", "authors": ["Qizhao Chen"], "year": 2025, "published": "2025-08-22T13:34:09Z", "updated": "2026-03-04T03:28:45Z", "arxiv_id": "http://arxiv.org/abs/2508.16378v2", "url": "https://arxiv.org/abs/2508.16378v2", "primary_category": "cs.CE", "categories": ["cs.CE", "q-fin.ST"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["Gemini large language model. These signals are incorporated into expected return estimates and used in a constrained mean-variance optimization framework. Backtesting across multiple cryptocurrencies shows that the integrated approach outperforms traditional benchmarks"], "key_results": ["Backtesting across multiple cryptocurrencies shows that the integrated approach outperforms traditional benchmarks, including momentum strategy, Bitcoin Long-Short strategy, and an equal-weighted portfolio, achieving stronger risk-adjusted returns and more consistent cumulative growth.", "Furthermore, comparing the sentiment-only and technical-only strategies shows that incorporating sentiment information alongside technical indicators can lead to more consistent performance gains.", "However, the strategies exhibit substantial drawdowns that coincide with known periods of market stress, indicating that additional risk-management components are required to improve stability."], "abstract": "Cryptocurrency markets are highly volatile and influenced by both price trends and market sentiment, making effective portfolio management challenging. This paper proposes a dynamic cryptocurrency portfolio strategy that integrates technical indicators and sentiment analysis to enhance investment decision-making. Market momentum is captured using the 14-day Relative Strength Index (RSI) and Simple Moving Average (SMA), while sentiment signals are extracted from news articles with VADER and further validated using the Google Gemini large language model. These signals are incorporated into expected return estimates and used in a constrained mean-variance optimization framework. Backtesting across multiple cryptocurrencies shows that the integrated approach outperforms traditional benchmarks, including momentum strategy, Bitcoin Long-Short strategy, and an equal-weighted portfolio, achieving stronger risk-adjusted returns and more consistent cumulative growth. Furthermore, comparing the sentiment-only and technical-only strategies shows that incorporating sentiment information alongside technical indicators can lead to more consistent performance gains. However, the strategies exhibit substantial drawdowns that coincide with known periods of market stress, indicating that additional risk-management components are required to improve stability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c1e811a24208236d", "title": "FEST: A Unified Framework for Evaluating Synthetic Tabular Data", "authors": ["Weijie Niu", "Alberto Huertas Celdran", "Karoline Siarsky", "Burkhard Stiller"], "year": 2025, "published": "2025-08-22T09:38:02Z", "updated": "2025-08-22T09:38:02Z", "arxiv_id": "http://arxiv.org/abs/2508.16254v1", "url": "https://arxiv.org/abs/2508.16254v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Synthetic data generation, leveraging generative machine learning techniques, offers a promising approach to mitigating privacy concerns associated with real-world data usage."], "abstract": "Synthetic data generation, leveraging generative machine learning techniques, offers a promising approach to mitigating privacy concerns associated with real-world data usage. Synthetic data closely resembles real-world data while maintaining strong privacy guarantees. However, a comprehensive assessment framework is still missing in the evaluation of synthetic data generation, especially when considering the balance between privacy preservation and data utility in synthetic data. This research bridges this gap by proposing FEST, a systematic framework for evaluating synthetic tabular data. FEST integrates diverse privacy metrics (attack-based and distance-based), along with similarity and machine learning utility metrics, to provide a holistic assessment. We develop FEST as an open-source Python-based library and validate it on multiple datasets, demonstrating its effectiveness in analyzing the privacy-utility trade-off of different synthetic data generation models. The source code of FEST is available on Github.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-93f4bfd9eabca367", "title": "HePGA: A Heterogeneous Processing-in-Memory based GNN Training Accelerator", "authors": ["Chukwufumnanya Ogbogu", "Gaurav Narang", "Biresh Kumar Joardar", "Janardhan Rao Doppa", "Krishnendu Chakrabarty", "Partha Pratim Pande"], "year": 2025, "published": "2025-08-22T00:06:50Z", "updated": "2025-08-22T00:06:50Z", "arxiv_id": "http://arxiv.org/abs/2508.16011v1", "url": "https://arxiv.org/abs/2508.16011v1", "primary_category": "cs.ET", "categories": ["cs.ET", "cs.AR", "cs.LG"], "pillar": "Quant × AI", "methods": ["Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Our experimental analysis shows that HePGA outperforms existing PIM-based architectures by up to 3.8x and 6.8x in energy-efficiency (TOPS/W) and compute efficiency (TOPS/mm2) respectively, without sacrificing the GNN prediction accuracy."], "abstract": "Processing-In-Memory (PIM) architectures offer a promising approach to accelerate Graph Neural Network (GNN) training and inference. However, various PIM devices such as ReRAM, FeFET, PCM, MRAM, and SRAM exist, with each device offering unique trade-offs in terms of power, latency, area, and non-idealities. A heterogeneous manycore architecture enabled by 3D integration can combine multiple PIM devices on a single platform, to enable energy-efficient and high-performance GNN training. In this work, we propose a 3D heterogeneous PIM-based accelerator for GNN training referred to as HePGA. We leverage the unique characteristics of GNN layers and associated computing kernels to optimize their mapping on to different PIM devices as well as planar tiers. Our experimental analysis shows that HePGA outperforms existing PIM-based architectures by up to 3.8x and 6.8x in energy-efficiency (TOPS/W) and compute efficiency (TOPS/mm2) respectively, without sacrificing the GNN prediction accuracy. Finally, we demonstrate the applicability of HePGA to accelerate inferencing of emerging transformer models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ab113676d5624cba", "title": "Advancing rail safety: An onboard measurement system of rolling stock wheel flange wear based on dynamic machine learning algorithms", "authors": ["Celestin Nkundineza", "James Ndodana Njaji", "Samrawit Abubeker", "Omar Gatera", "Damien Hanyurwimfura"], "year": 2025, "published": "2025-08-21T21:12:48Z", "updated": "2025-08-21T21:12:48Z", "arxiv_id": "http://arxiv.org/abs/2508.15963v1", "url": "https://arxiv.org/abs/2508.15963v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CE", "eess.SP", "eess.SY", "physics.ins-det"], "pillar": "Quant × AI", "methods": ["Safety / Jailbreak", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["The results show that the dynamic machine learning algorithm effectively counter sensor nonlinear response to temperature effects, achieving an accuracy of 96.5 %, with a minimal runtime."], "abstract": "Rail and wheel interaction functionality is pivotal to the railway system safety, requiring accurate measurement systems for optimal safety monitoring operation. This paper introduces an innovative onboard measurement system for monitoring wheel flange wear depth, utilizing displacement and temperature sensors. Laboratory experiments are conducted to emulate wheel flange wear depth and surrounding temperature fluctuations in different periods of time. Employing collected data, the training of machine learning algorithms that are based on regression models, is dynamically automated. Further experimentation results, using standards procedures, validate the system's efficacy. To enhance accuracy, an infinite impulse response filter (IIR) that mitigates vehicle dynamics and sensor noise is designed. Filter parameters were computed based on specifications derived from a Fast Fourier Transform analysis of locomotive simulations and emulation experiments data. The results show that the dynamic machine learning algorithm effectively counter sensor nonlinear response to temperature effects, achieving an accuracy of 96.5 %, with a minimal runtime. The real-time noise reduction via IIR filter enhances the accuracy up to 98.2 %. Integrated with railway communication embedded systems such as Internet of Things devices, this advanced monitoring system offers unparalleled real-time insights into wheel flange wear and track irregular conditions that cause it, ensuring heightened safety and efficiency in railway systems operations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0ba5762d6001b9f9", "title": "Industrial Steel Slag Flow Data Loading Method for Deep Learning Applications", "authors": ["Mert Sehri", "Ana Cardoso", "Francisco de Assis Boldt", "Patrick Dumond"], "year": 2025, "published": "2025-08-21T20:48:11Z", "updated": "2025-08-21T20:48:11Z", "arxiv_id": "http://arxiv.org/abs/2509.00034v1", "url": "https://arxiv.org/abs/2509.00034v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["A hybrid deep learning model combining one-dimensional convolutional neural networks and long short-term memory layers is implemented, tested, and benchmarked against a standard one-dimensional convolutional neural network.", "Results show that the hybrid convolutional neural network and long short-term memory architecture, when combined with root mean square preprocessing and a selective embedding data loading strategy, achieves robust classification accuracy, outperforming traditional models and loading techniques.", "This work presents a practical and scalable solution for real-time slag flow monitoring, contributing to improved reliability and operational efficiency in steel manufacturing."], "abstract": "Steel casting processes are vulnerable to financial losses due to slag flow contamination, making accurate slag flow condition detection essential. This study introduces a novel cross-domain diagnostic method using vibration data collected from an industrial steel foundry to identify various stages of slag flow. A hybrid deep learning model combining one-dimensional convolutional neural networks and long short-term memory layers is implemented, tested, and benchmarked against a standard one-dimensional convolutional neural network. The proposed method processes raw time-domain vibration signals from accelerometers and evaluates performance across 16 distinct domains using a realistic cross-domain dataset split. Results show that the hybrid convolutional neural network and long short-term memory architecture, when combined with root mean square preprocessing and a selective embedding data loading strategy, achieves robust classification accuracy, outperforming traditional models and loading techniques. The highest test accuracy of 99.10 +/- 0.30 demonstrates the method's capability for generalization and industrial relevance. This work presents a practical and scalable solution for real-time slag flow monitoring, contributing to improved reliability and operational efficiency in steel manufacturing.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b3fe34c71e2367ab", "title": "Probabilistic Forecasting Cryptocurrencies Volatility: From Point to Quantile Forecasts", "authors": ["Grzegorz Dudek", "Witold Orzeszko", "Piotr Fiszeder"], "year": 2025, "published": "2025-08-21T18:42:11Z", "updated": "2025-08-21T18:42:11Z", "arxiv_id": "http://arxiv.org/abs/2508.15922v1", "url": "https://arxiv.org/abs/2508.15922v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": ["Deep Research", "Quant / Trading"], "benchmarks": [], "models": ["phisticated"], "key_results": ["Our empirical results for Bitcoin demonstrate that the Quantile Estimation through Residual Simulation (QRS) method, particularly when applied to linear base models operating on log-transformed realized volatility data, consistently outperforms more sophisticated alternatives."], "abstract": "Cryptocurrency markets are characterized by extreme volatility, making accurate forecasts essential for effective risk management and informed trading strategies. Traditional deterministic (point) forecasting methods are inadequate for capturing the full spectrum of potential volatility outcomes, underscoring the importance of probabilistic approaches. To address this limitation, this paper introduces probabilistic forecasting methods that leverage point forecasts from a wide range of base models, including statistical (HAR, GARCH, ARFIMA) and machine learning (e.g. LASSO, SVR, MLP, Random Forest, LSTM) algorithms, to estimate conditional quantiles of cryptocurrency realized variance. To the best of our knowledge, this is the first study in the literature to propose and systematically evaluate probabilistic forecasts of variance in cryptocurrency markets based on predictions derived from multiple base models. Our empirical results for Bitcoin demonstrate that the Quantile Estimation through Residual Simulation (QRS) method, particularly when applied to linear base models operating on log-transformed realized volatility data, consistently outperforms more sophisticated alternatives. Additionally, we highlight the robustness of the probabilistic stacking framework, providing comprehensive insights into uncertainty and risk inherent in cryptocurrency volatility forecasting. This research fills a significant gap in the literature, contributing practical probabilistic forecasting methodologies tailored specifically to cryptocurrency markets.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-f14de040c16e709b", "title": "Non-parametric Causal Discovery for EU Allowances Returns Through the Information Imbalance", "authors": ["Cristiano Salvagnin", "Vittorio del Tatto", "Maria Elena De Giuli", "Antonietta Mira", "Aldo Glielmo"], "year": 2025, "published": "2025-08-21T15:52:14Z", "updated": "2025-08-21T15:52:14Z", "arxiv_id": "http://arxiv.org/abs/2508.15667v1", "url": "https://arxiv.org/abs/2508.15667v1", "primary_category": "q-fin.CP", "categories": ["q-fin.CP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["We propose to use a recently introduced non-parametric tool named Differentiable Information Imbalance (DII) to identify variables that are causally related -- potentially through non-linear relationships -- to the financial returns of the European Union Allowances (EUAs) within the EU Emissions Trading System (EU ETS)."], "abstract": "We propose to use a recently introduced non-parametric tool named Differentiable Information Imbalance (DII) to identify variables that are causally related -- potentially through non-linear relationships -- to the financial returns of the European Union Allowances (EUAs) within the EU Emissions Trading System (EU ETS). We examine data from January 2013 to April 2024 and compare the DII approach with multivariate Granger causality, a well-known linear approach based on VAR models. We find significant overlap among the causal variables identified by linear and non-linear methods, such as the coal futures prices and the IBEX35 index. We also find important differences between the two causal sets identified. On two synthetic datasets, we show how these differences could originate from limitations of the linear methodology.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0bba6bc0e140dd35", "title": "Stabilization of Perturbed Loss Function: Differential Privacy without Gradient Noise", "authors": ["Salman Habib", "Remi Chou", "Taejoon Kim"], "year": 2025, "published": "2025-08-21T12:54:19Z", "updated": "2025-08-21T12:54:19Z", "arxiv_id": "http://arxiv.org/abs/2508.15523v1", "url": "https://arxiv.org/abs/2508.15523v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.IT"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Unlike gradient-based mechanisms such as differentially private stochastic gradient descent (DP-SGD), SPOF does not require injecting noise into the gradients of the loss function, which improves both computational efficiency and stability.", "Our results show that SPOF achieves, on average, up to 3.5% higher reconstruction accuracy and reduces mean training time by up to 57.2% compared to DP-SGD, demonstrating superior privacy-utility trade-offs in multi-user environments."], "abstract": "We propose SPOF (Stabilization of Perturbed Loss Function), a differentially private training mechanism intended for multi-user local differential privacy (LDP). SPOF perturbs a stabilized Taylor expanded polynomial approximation of a model's training loss function, where each user's data is privatized by calibrated noise added to the coefficients of the polynomial. Unlike gradient-based mechanisms such as differentially private stochastic gradient descent (DP-SGD), SPOF does not require injecting noise into the gradients of the loss function, which improves both computational efficiency and stability. This formulation naturally supports simultaneous privacy guarantees across all users. Moreover, SPOF exhibits robustness to environmental noise during training, maintaining stable performance even when user inputs are corrupted. We compare SPOF with a multi-user extension of DP-SGD, evaluating both methods in a wireless body area network (WBAN) scenario involving heterogeneous user data and stochastic channel noise from body sensors. Our results show that SPOF achieves, on average, up to 3.5% higher reconstruction accuracy and reduces mean training time by up to 57.2% compared to DP-SGD, demonstrating superior privacy-utility trade-offs in multi-user environments.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d0034c9c62c347e2", "title": "Enhancing Forecasting with a 2D Time Series Approach for Cohort-Based Data", "authors": ["Yonathan Guttel", "Orit Moradov", "Nachi Lieder", "Asnat Greenstein-Messica"], "year": 2025, "published": "2025-08-21T08:53:40Z", "updated": "2025-08-21T08:53:40Z", "arxiv_id": "http://arxiv.org/abs/2508.15369v1", "url": "https://arxiv.org/abs/2508.15369v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["This paper introduces a novel two-dimensional (2D) time series forecasting model that integrates cohort behavior over time, addressing challenges in small data environments."], "abstract": "This paper introduces a novel two-dimensional (2D) time series forecasting model that integrates cohort behavior over time, addressing challenges in small data environments. We demonstrate its efficacy using multiple real-world datasets, showcasing superior performance in accuracy and adaptability compared to reference models. The approach offers valuable insights for strategic decision-making across industries facing financial and marketing forecasting challenges.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-19b0b41797e24920", "title": "Locally Pareto-Optimal Interpretations for Black-Box Machine Learning Models", "authors": ["Aniruddha Joshi", "Supratik Chakraborty", "S Akshay", "Shetal Shah", "Hazem Torfah", "Sanjit Seshia"], "year": 2025, "published": "2025-08-21T04:11:20Z", "updated": "2025-08-21T04:11:20Z", "arxiv_id": "http://arxiv.org/abs/2508.15220v1", "url": "https://arxiv.org/abs/2508.15220v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.LO"], "pillar": "Quant × AI", "methods": ["Search / MCTS"], "benchmarks": [], "models": ["yields"], "key_results": ["We demonstrate the efficacy of our approach on a set of benchmarks, comparing it against previous methods for exploring the Pareto-optimal front of interpretations."], "abstract": "Creating meaningful interpretations for black-box machine learning models involves balancing two often conflicting objectives: accuracy and explainability. Exploring the trade-off between these objectives is essential for developing trustworthy interpretations. While many techniques for multi-objective interpretation synthesis have been developed, they typically lack formal guarantees on the Pareto-optimality of the results. Methods that do provide such guarantees, on the other hand, often face severe scalability limitations when exploring the Pareto-optimal space. To address this, we develop a framework based on local optimality guarantees that enables more scalable synthesis of interpretations. Specifically, we consider the problem of synthesizing a set of Pareto-optimal interpretations with local optimality guarantees, within the immediate neighborhood of each solution. Our approach begins with a multi-objective learning or search technique, such as Multi-Objective Monte Carlo Tree Search, to generate a best-effort set of Pareto-optimal candidates with respect to accuracy and explainability. We then verify local optimality for each candidate as a Boolean satisfiability problem, which we solve using a SAT solver. We demonstrate the efficacy of our approach on a set of benchmarks, comparing it against previous methods for exploring the Pareto-optimal front of interpretations. In particular, we show that our approach yields interpretations that closely match those synthesized by methods offering global guarantees.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-103e81654d20998b", "title": "Kernel-based Equalized Odds: A Quantification of Accuracy-Fairness Trade-off in Fair Representation Learning", "authors": ["Yijin Ni", "Xiaoming Huo"], "year": 2025, "published": "2025-08-20T21:41:34Z", "updated": "2025-08-20T21:41:34Z", "arxiv_id": "http://arxiv.org/abs/2508.15084v1", "url": "https://arxiv.org/abs/2508.15084v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "Quant × AI", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["This paper introduces a novel kernel-based formulation of the Equalized Odds (EO) criterion, denoted as $EO_k$, for fair representation learning (FRL) in supervised settings."], "abstract": "This paper introduces a novel kernel-based formulation of the Equalized Odds (EO) criterion, denoted as $EO_k$, for fair representation learning (FRL) in supervised settings. The central goal of FRL is to mitigate discrimination regarding a sensitive attribute $S$ while preserving prediction accuracy for the target variable $Y$. Our proposed criterion enables a rigorous and interpretable quantification of three core fairness objectives: independence (prediction $\\hat{Y}$ is independent of $S$), separation (also known as equalized odds; prediction $\\hat{Y}$ is independent with $S$ conditioned on target attribute $Y$), and calibration ($Y$ is independent of $S$ conditioned on the prediction $\\hat{Y}$). Under both unbiased ($Y$ is independent of $S$) and biased ($Y$ depends on $S$) conditions, we show that $EO_k$ satisfies both independence and separation in the former, and uniquely preserves predictive accuracy while lower bounding independence and calibration in the latter, thereby offering a unified analytical characterization of the tradeoffs among these fairness criteria. We further define the empirical counterpart, $\\hat{EO}_k$, a kernel-based statistic that can be computed in quadratic time, with linear-time approximations also available. A concentration inequality for $\\hat{EO}_k$ is derived, providing performance guarantees and error bounds, which serve as practical certificates of fairness compliance. While our focus is on theoretical development, the results lay essential groundwork for principled and provably fair algorithmic design in future empirical studies.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-175ec458af8f3153", "title": "A Systematic Survey of Model Extraction Attacks and Defenses: State-of-the-Art and Perspectives", "authors": ["Kaixiang Zhao", "Lincan Li", "Kaize Ding", "Neil Zhenqiang Gong", "Yue Zhao", "Yushun Dong"], "year": 2025, "published": "2025-08-20T19:49:59Z", "updated": "2025-08-27T05:10:26Z", "arxiv_id": "http://arxiv.org/abs/2508.15031v2", "url": "https://arxiv.org/abs/2508.15031v2", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": ["Deep Research"], "benchmarks": [], "models": ["phisticated"], "key_results": ["Machine learning (ML) models have significantly grown in complexity and utility, driving advances across multiple domains."], "abstract": "Machine learning (ML) models have significantly grown in complexity and utility, driving advances across multiple domains. However, substantial computational resources and specialized expertise have historically restricted their wide adoption. Machine-Learning-as-a-Service (MLaaS) platforms have addressed these barriers by providing scalable, convenient, and affordable access to sophisticated ML models through user-friendly APIs. While this accessibility promotes widespread use of advanced ML capabilities, it also introduces vulnerabilities exploited through Model Extraction Attacks (MEAs). Recent studies have demonstrated that adversaries can systematically replicate a target model's functionality by interacting with publicly exposed interfaces, posing threats to intellectual property, privacy, and system security. In this paper, we offer a comprehensive survey of MEAs and corresponding defense strategies. We propose a novel taxonomy that classifies MEAs according to attack mechanisms, defense approaches, and computing environments. Our analysis covers various attack techniques, evaluates their effectiveness, and highlights challenges faced by existing defenses, particularly the critical trade-off between preserving model utility and ensuring security. We further assess MEAs within different computing paradigms and discuss their technical, ethical, legal, and societal implications, along with promising directions for future research. This systematic survey aims to serve as a valuable reference for researchers, practitioners, and policymakers engaged in AI security and privacy. Additionally, we maintain an online repository continuously updated with related literature at https://github.com/kzhao5/ModelExtractionPapers.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-936bf2490f992eb9", "title": "Neural Network Quantization for Microcontrollers: A Comprehensive Survey of Methods, Platforms, and Applications", "authors": ["Hamza A. Abushahla", "Dara Varam", "Ariel Justine N. Panopio", "Mohamed I. AlHajri"], "year": 2025, "published": "2025-08-20T18:56:26Z", "updated": "2026-01-07T11:09:58Z", "arxiv_id": "http://arxiv.org/abs/2508.15008v4", "url": "https://arxiv.org/abs/2508.15008v4", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.AR"], "pillar": "Quant × AI", "methods": ["Quantization", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["The deployment of Quantized Neural Networks (QNNs) on resource-constrained edge devices, such as microcontrollers (MCUs), introduces fundamental challenges in balancing model performance, computational complexity, and memory constraints."], "abstract": "The deployment of Quantized Neural Networks (QNNs) on resource-constrained edge devices, such as microcontrollers (MCUs), introduces fundamental challenges in balancing model performance, computational complexity, and memory constraints. Tiny Machine Learning (TinyML) addresses these issues by jointly advancing machine learning algorithms, hardware architectures, and software optimization techniques to enable deep neural network inference on embedded systems. This survey provides a hardware-oriented perspective on neural network quantization, systematically reviewing the quantization methods most relevant to MCUs and extreme-edge devices. Particular emphasis is placed on the critical trade-offs between model performance and the capabilities of MCU-class hardware, including memory hierarchies, numerical representations, and accelerator support. The survey further reviews contemporary MCU hardware platforms, including ARM-based and RISC-V-based designs, as well as MCUs integrating neural processing units (NPUs) for low-precision inference, together with the supporting software stacks. In addition, we analyze real-world deployments of quantized models on MCUs and consolidate the application domains in which such systems are used. Finally, we discuss open challenges and outline promising future directions toward scalable, energy-efficient, and sustainable AI deployment on edge devices.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ddfd604d9da23e48", "title": "Graph Learning for Foreign Exchange Rate Prediction and Statistical Arbitrage", "authors": ["Yoonsik Hong", "Diego Klabjan"], "year": 2025, "published": "2025-08-20T15:29:31Z", "updated": "2025-08-20T15:29:31Z", "arxiv_id": "http://arxiv.org/abs/2508.14784v1", "url": "https://arxiv.org/abs/2508.14784v1", "primary_category": "q-fin.TR", "categories": ["q-fin.TR"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": ["yields"], "key_results": ["The experimental results demonstrate that our FXRP method yields statistically significant improvements in mean squared error, and that the FXSA method achieves a 61.89% higher information ratio and a 45.51% higher Sortino ratio than a benchmark."], "abstract": "We propose a two-step graph learning approach for foreign exchange statistical arbitrages (FXSAs), addressing two key gaps in prior studies: the absence of graph-learning methods for foreign exchange rate prediction (FXRP) that leverage multi-currency and currency-interest rate relationships, and the disregard of the time lag between price observation and trade execution. In the first step, to capture complex multi-currency and currency-interest rate relationships, we formulate FXRP as an edge-level regression problem on a discrete-time spatiotemporal graph. This graph consists of currencies as nodes and exchanges as edges, with interest rates and foreign exchange rates serving as node and edge features, respectively. We then introduce a graph-learning method that leverages the spatiotemporal graph to address the FXRP problem. In the second step, we present a stochastic optimization problem to exploit FXSAs while accounting for the observation-execution time lag. To address this problem, we propose a graph-learning method that enforces constraints through projection and ReLU, maximizes risk-adjusted return by leveraging a graph with exchanges as nodes and influence relationships as edges, and utilizes the predictions from the FXRP method for the constraint parameters and node features. Moreover, we prove that our FXSA method satisfies empirical arbitrage constraints. The experimental results demonstrate that our FXRP method yields statistically significant improvements in mean squared error, and that the FXSA method achieves a 61.89% higher information ratio and a 45.51% higher Sortino ratio than a benchmark. Our approach provides a novel perspective on FXRP and FXSA within the context of graph learning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-66e9566a267e1695", "title": "XFinBench: Benchmarking LLMs in Complex Financial Problem Solving and Reasoning", "authors": ["Zhihan Zhang", "Yixin Cao", "Lizi Liao"], "year": 2025, "published": "2025-08-20T15:23:35Z", "updated": "2025-08-20T15:23:35Z", "arxiv_id": "http://arxiv.org/abs/2508.15861v1", "url": "https://arxiv.org/abs/2508.15861v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "Quant × AI", "methods": ["VLM", "Quant / Trading"], "benchmarks": [], "models": ["o1"], "key_results": ["We further construct a knowledge bank with 3,032 finance terms for knowledge augmentation analysis, and find that relevant knowledge to the question only brings consistent accuracy improvements to small open-source model."], "abstract": "Solving financial problems demands complex reasoning, multimodal data processing, and a broad technical understanding, presenting unique challenges for current large language models (LLMs). We introduce XFinBench, a novel benchmark with 4,235 examples designed to evaluate LLM's ability in solving complex, knowledge-intensive financial problems across diverse graduate-level finance topics with multi-modal context. We identify five core capabilities of LLMs using XFinBench, i.e, terminology understanding, temporal reasoning, future forecasting, scenario planning, and numerical modelling. Upon XFinBench, we conduct extensive experiments on 18 leading models. The result shows that o1 is the best-performing text-only model with an overall accuracy of 67.3%, but still lags significantly behind human experts with 12.5%, especially in temporal reasoning and scenario planning capabilities. We further construct a knowledge bank with 3,032 finance terms for knowledge augmentation analysis, and find that relevant knowledge to the question only brings consistent accuracy improvements to small open-source model. Additionally, our error analysis reveals that rounding errors during calculation and blindness to position and intersection of curves in the image are two primary issues leading to model's poor performance in calculating and visual-context questions, respectively. Code and dataset are accessible via GitHub: https://github.com/Zhihan72/XFinBench.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d38895c7889653d6", "title": "AFABench: A Generic Framework for Benchmarking Active Feature Acquisition", "authors": ["Valter Schütz", "Han Wu", "Reza Rezvan", "Linus Aronsson", "Morteza Haghir Chehreghani"], "year": 2025, "published": "2025-08-20T14:29:16Z", "updated": "2026-02-22T20:33:13Z", "arxiv_id": "http://arxiv.org/abs/2508.14734v3", "url": "https://arxiv.org/abs/2508.14734v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Search / MCTS", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Active Feature Acquisition (AFA) addresses this challenge by dynamically selecting a subset of informative features for each data instance, trading predictive performance against acquisition cost."], "abstract": "In many real-world scenarios, acquiring all features of a data instance can be expensive or impractical due to monetary cost, latency, or privacy concerns. Active Feature Acquisition (AFA) addresses this challenge by dynamically selecting a subset of informative features for each data instance, trading predictive performance against acquisition cost. While numerous methods have been proposed for AFA, ranging from myopic information-theoretic strategies to non-myopic reinforcement learning approaches, fair and systematic evaluation of these methods has been hindered by a lack of standardized benchmarks. In this paper, we introduce AFABench, the first benchmark framework for AFA. Our benchmark includes a diverse set of synthetic and real-world datasets, supports a wide range of acquisition policies, and provides a modular design that enables easy integration of new methods and tasks. We implement and evaluate representative algorithms from all major categories, including static, myopic, and reinforcement learning-based approaches. To test the lookahead capabilities of AFA policies, we introduce a novel synthetic dataset, CUBE-NM, designed to expose the limitations of myopic selection. Our results highlight key trade-offs between different AFA strategies and provide actionable insights for future research. The benchmark code is available at: https://github.com/Linusaronsson/AFA-Benchmark.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-08fc03436938f71f", "title": "Deep Learning for Short Term Equity Trend Forecasting: A Behavior Driven Multi Factor Approach", "authors": ["Yuqi Luan"], "year": 2025, "published": "2025-08-20T12:15:32Z", "updated": "2025-08-20T12:15:32Z", "arxiv_id": "http://arxiv.org/abs/2508.14656v1", "url": "https://arxiv.org/abs/2508.14656v1", "primary_category": "q-fin.TR", "categories": ["q-fin.TR"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Empirical evaluation demonstrates that the dual-task MLP achieves superior and stable performance across both predictive accuracy and economic relevance, as measured by information coefficient (IC), information ratio (IR), and portfolio backtesting results.", "Comparative experiments further show that deep learning methods outperform linear baselines by effectively capturing structural interactions between factors."], "abstract": "This study proposes a behaviorally-informed multi-factor stock selection framework that integrates short-cycle technical alpha signals with deep learning. We design a dual-task multilayer perceptron (MLP) that jointly predicts five-day future returns and directional price movements, thereby capturing nonlinear market behaviors such as volume-price divergence, momentum-driven herding, and bottom reversals. The model is trained on 40 carefully constructed factors derived from price-volume patterns and behavioral finance insights. Empirical evaluation demonstrates that the dual-task MLP achieves superior and stable performance across both predictive accuracy and economic relevance, as measured by information coefficient (IC), information ratio (IR), and portfolio backtesting results. Comparative experiments further show that deep learning methods outperform linear baselines by effectively capturing structural interactions between factors. This work highlights the potential of structure-aware deep learning in enhancing multi-factor modeling and provides a practical framework for short-horizon quantitative investment strategies.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-aec7f943b8458fe5", "title": "Directional Price Forecasting in the Continuous Intraday Market under Consideration of Neighboring Products and Limit Order Books", "authors": ["Timothée Hornek", "Sergio Potenciano Menci", "Ivan Pavić"], "year": 2025, "published": "2025-08-20T09:52:14Z", "updated": "2025-08-20T09:52:14Z", "arxiv_id": "http://arxiv.org/abs/2509.04452v1", "url": "https://arxiv.org/abs/2509.04452v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST"], "pillar": "Quant × AI", "methods": ["Deep Research", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["In addition, features from neighboring products; especially those with delivery start times that overlap with the trading period of the target product; improve forecast accuracy."], "abstract": "The increasing penetration of variable renewable energy and flexible demand technologies, such as electric vehicles and heat pumps, introduces significant uncertainty in power systems, resulting in greater imbalance; defined as the deviation between scheduled and actual supply or demand. Short-term power markets, such as the European continuous intraday market, play a critical role in mitigating these imbalances by enabling traders to adjust forecasts close to real time. Due to the high volatility of the continuous intraday market, traders increasingly rely on electricity price forecasting to guide trading decisions and mitigate price risk. However most electricity price forecasting approaches in the literature simplify the forecasting task. They focus on single benchmark prices, neglecting intra-product price dynamics and price signals from the limit order book. They also underuse high-frequency and cross-product price data.\n  In turn, we propose a novel directional electricity price forecasting method for hourly products in the European continuous intraday market. Our method incorporates short-term features from both hourly and quarter-hourly products and is evaluated using German European Power Exchange data from 2024-2025. The results indicate that features derived from the limit order book are the most influential exogenous variables. In addition, features from neighboring products; especially those with delivery start times that overlap with the trading period of the target product; improve forecast accuracy. Finally, our evaluation of the value captured by our electricity price forecasting suggests that the proposed electricity price forecasting method has the potential to generate profit when applied in trading strategies.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-546ae684c178b7e3", "title": "Beyond ReLU: Chebyshev-DQN for Enhanced Deep Q-Networks", "authors": ["Saman Yazdannik", "Morteza Tayefi", "Shamim Sanisales"], "year": 2025, "published": "2025-08-20T08:41:15Z", "updated": "2025-08-20T08:41:15Z", "arxiv_id": "http://arxiv.org/abs/2508.14536v1", "url": "https://arxiv.org/abs/2508.14536v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["ying"], "key_results": ["By leveraging the powerful function approximation properties of Chebyshev polynomials, we hypothesize that the Ch-DQN can learn more efficiently and achieve higher performance.", "We evaluate our proposed model on the CartPole-v1 benchmark and compare it against a standard DQN with a comparable number of parameters.", "Our results demonstrate that the Ch-DQN with a moderate polynomial degree (N=4) achieves significantly better asymptotic performance, outperforming the baseline by approximately 39\\%."], "abstract": "The performance of Deep Q-Networks (DQN) is critically dependent on the ability of its underlying neural network to accurately approximate the action-value function. Standard function approximators, such as multi-layer perceptrons, may struggle to efficiently represent the complex value landscapes inherent in many reinforcement learning problems. This paper introduces a novel architecture, the Chebyshev-DQN (Ch-DQN), which integrates a Chebyshev polynomial basis into the DQN framework to create a more effective feature representation. By leveraging the powerful function approximation properties of Chebyshev polynomials, we hypothesize that the Ch-DQN can learn more efficiently and achieve higher performance. We evaluate our proposed model on the CartPole-v1 benchmark and compare it against a standard DQN with a comparable number of parameters. Our results demonstrate that the Ch-DQN with a moderate polynomial degree (N=4) achieves significantly better asymptotic performance, outperforming the baseline by approximately 39\\%. However, we also find that the choice of polynomial degree is a critical hyperparameter, as a high degree (N=8) can be detrimental to learning. This work validates the potential of using orthogonal polynomial bases in deep reinforcement learning while also highlighting the trade-offs involved in model complexity.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e177ea842101f6ef", "title": "DuPO: Enabling Reliable LLM Self-Verification via Dual Preference Optimization", "authors": ["Shuaijie She", "Yu Bao", "Yu Lu", "Lu Xu", "Tao Li", "Wenhao Zhu", "Shujian Huang", "Shanbo Cheng", "Lu Lu", "Yuxuan Wang"], "year": 2025, "published": "2025-08-20T06:31:18Z", "updated": "2025-08-20T06:31:18Z", "arxiv_id": "http://arxiv.org/abs/2508.14460v1", "url": "https://arxiv.org/abs/2508.14460v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Test-time Scaling", "Preference Optimization", "Quant / Trading"], "benchmarks": ["MATH"], "models": [], "key_results": ["Empirically, DuPO achieves substantial gains across diverse tasks: it enhances the average translation quality by 2.13 COMET over 756 directions, boosts the mathematical reasoning accuracy by an average of 6.4 points on three challenge benchmarks, and enhances performance by 9.3 points as an inference-time reranker (trading computation for accuracy)."], "abstract": "We present DuPO, a dual learning-based preference optimization framework that generates annotation-free feedback via a generalized duality. DuPO addresses two key limitations: Reinforcement Learning with Verifiable Rewards (RLVR)'s reliance on costly labels and applicability restricted to verifiable tasks, and traditional dual learning's restriction to strictly dual task pairs (e.g., translation and back-translation). Specifically, DuPO decomposes a primal task's input into known and unknown components, then constructs its dual task to reconstruct the unknown part using the primal output and known information (e.g., reversing math solutions to recover hidden variables), broadening applicability to non-invertible tasks. The quality of this reconstruction serves as a self-supervised reward to optimize the primal task, synergizing with LLMs' ability to instantiate both tasks via a single model. Empirically, DuPO achieves substantial gains across diverse tasks: it enhances the average translation quality by 2.13 COMET over 756 directions, boosts the mathematical reasoning accuracy by an average of 6.4 points on three challenge benchmarks, and enhances performance by 9.3 points as an inference-time reranker (trading computation for accuracy). These results position DuPO as a scalable, general, and annotation-free paradigm for LLM optimization.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-cf9e7e577758ce80", "title": "Understanding Practitioners Perspectives on Monitoring Machine Learning Systems", "authors": ["Hira Naveed", "John Grundy", "Chetan Arora", "Hourieh Khalajzadeh", "Omar Haggag"], "year": 2025, "published": "2025-08-20T00:58:35Z", "updated": "2025-08-20T00:58:35Z", "arxiv_id": "http://arxiv.org/abs/2509.25195v1", "url": "https://arxiv.org/abs/2509.25195v1", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["This paper explores the strategies, challenges, and improvement opportunities for monitoring ML systems from the practitioners perspective.", "The desired improvements from the practitioners perspective are: automated generation and deployment of monitors, improved support for performance and fairness monitoring, and recommendations for resolving runtime issues."], "abstract": "Given the inherent non-deterministic nature of machine learning (ML) systems, their behavior in production environments can lead to unforeseen and potentially dangerous outcomes. For a timely detection of unwanted behavior and to prevent organizations from financial and reputational damage, monitoring these systems is essential. This paper explores the strategies, challenges, and improvement opportunities for monitoring ML systems from the practitioners perspective. We conducted a global survey of 91 ML practitioners to collect diverse insights into current monitoring practices for ML systems. We aim to complement existing research through our qualitative and quantitative analyses, focusing on prevalent runtime issues, industrial monitoring and mitigation practices, key challenges, and desired enhancements in future monitoring tools. Our findings reveal that practitioners frequently struggle with runtime issues related to declining model performance, exceeding latency, and security violations. While most prefer automated monitoring for its increased efficiency, many still rely on manual approaches due to the complexity or lack of appropriate automation solutions. Practitioners report that the initial setup and configuration of monitoring tools is often complicated and challenging, particularly when integrating with ML systems and setting alert thresholds. Moreover, practitioners find that monitoring adds extra workload, strains resources, and causes alert fatigue. The desired improvements from the practitioners perspective are: automated generation and deployment of monitors, improved support for performance and fairness monitoring, and recommendations for resolving runtime issues. These insights offer valuable guidance for the future development of ML monitoring tools that are better aligned with practitioners needs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7e45bb0827d5c6b4", "title": "Formal Algorithms for Model Efficiency", "authors": ["Naman Tyagi", "Srishti Das", "Kunal", "Vatsal Gupta"], "year": 2025, "published": "2025-08-19T16:54:02Z", "updated": "2025-08-19T16:54:02Z", "arxiv_id": "http://arxiv.org/abs/2508.14000v1", "url": "https://arxiv.org/abs/2508.14000v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Distillation", "Quantization"], "benchmarks": [], "models": ["ying"], "key_results": ["We introduce the Knob-Meter-Rule (KMR) framework, a unified formalism for representing and reasoning about model efficiency techniques in deep learning."], "abstract": "We introduce the Knob-Meter-Rule (KMR) framework, a unified formalism for representing and reasoning about model efficiency techniques in deep learning. By abstracting diverse methods, including pruning, quantization, knowledge distillation, and parameter-efficient architectures, into a consistent set of controllable knobs, deterministic rules, and measurable meters, KMR provides a mathematically precise and modular perspective on efficiency optimization. The framework enables systematic composition of multiple techniques, flexible policy-driven application, and iterative budgeted optimization through the Budgeted-KMR algorithm. We demonstrate how well-known efficiency methods can be instantiated as KMR triples and present concise algorithmic templates for each. The framework highlights underlying relationships between methods, facilitates hybrid pipelines, and lays the foundation for future research in automated policy learning, dynamic adaptation, and theoretical analysis of cost-quality trade-offs. Overall, KMR offers both a conceptual and practical tool for unifying and advancing model efficiency research.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-81132e66875564e3", "title": "Automated Energy-Aware Time-Series Model Deployment on Embedded FPGAs for Resilient Combined Sewer Overflow Management", "authors": ["Tianheng Ling", "Vipin Singh", "Chao Qian", "Felix Biessmann", "Gregor Schiele"], "year": 2025, "published": "2025-08-19T15:06:04Z", "updated": "2026-04-21T10:31:21Z", "arxiv_id": "http://arxiv.org/abs/2508.13905v2", "url": "https://arxiv.org/abs/2508.13905v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Quantization", "Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": ["yields"], "key_results": ["Evaluated on real-world sewer data, the selected 8-bit Transformer model, trained on 24 hours of historical measurements, achieves high accuracy (MSE 0.0376) at an energy cost of 0.370 mJ per inference."], "abstract": "Extreme weather events, intensified by climate change, increasingly challenge aging combined sewer systems, raising the risk of untreated wastewater overflow. Accurate forecasting of sewer overflow basin filling levels can provide actionable insights for early intervention, helping mitigating uncontrolled discharge. In recent years, AI-based forecasting methods have offered scalable alternatives to traditional physics-based models, but their reliance on cloud computing limits their reliability during communication outages. To address this, we propose an end-to-end forecasting framework that enables energy-efficient inference directly on edge devices. Our solution integrates lightweight Transformer and Long Short-Term Memory (LSTM) models, compressed via integer-only quantization for efficient on-device execution. Moreover, an automated hardware-aware deployment pipeline is used to search for optimal model configurations by jointly minimizing prediction error and energy consumption on an AMD Spartan-7 XC7S15 FPGA. Evaluated on real-world sewer data, the selected 8-bit Transformer model, trained on 24 hours of historical measurements, achieves high accuracy (MSE 0.0376) at an energy cost of 0.370 mJ per inference. In contrast, the optimal 8-bit LSTM model requires significantly less energy (0.009 mJ, over 40x lower) but yields 14.89% worse accuracy (MSE 0.0432) and much longer training time. This trade-off highlights the need to align model selection with deployment priorities, favoring LSTM for ultra-low energy consumption or Transformer for higher predictive accuracy. In general, our work enables local, energy-efficient forecasting, contributing to more resilient combined sewer systems. All code can be found in the GitHub Repository (https://github.com/tianheng-ling/EdgeOverflowForecast).", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-57b87229cd5fdafa", "title": "One-Step Flow Q-Learning: Addressing the Diffusion Policy Bottleneck in Offline Reinforcement Learning", "authors": ["Thanh Nguyen", "Chang D. Yoo"], "year": 2025, "published": "2025-08-19T15:05:55Z", "updated": "2026-02-24T10:29:51Z", "arxiv_id": "http://arxiv.org/abs/2508.13904v3", "url": "https://arxiv.org/abs/2508.13904v3", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Distillation", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Extensive experiments on the D4RL benchmark show that OFQL, despite generating actions in a single step, not only significantly reduces computation during both training and inference but also outperforms multi-step DQL by a large margin.", "Furthermore, OFQL surpasses all other baselines, achieving state-of-the-art performance in D4RL."], "abstract": "Diffusion Q-Learning (DQL) has established diffusion policies as a high-performing paradigm for offline reinforcement learning, but its reliance on multi-step denoising for action generation renders both training and inference slow and fragile. Existing efforts to accelerate DQL toward one-step denoising typically rely on auxiliary modules or policy distillation, sacrificing either simplicity or performance. It remains unclear whether a one-step policy can be trained directly without such trade-offs. To this end, we introduce One-Step Flow Q-Learning (OFQL), a novel framework that enables effective one-step action generation during both training and inference, without auxiliary modules or distillation. OFQL reformulates the DQL policy within the Flow Matching (FM) paradigm but departs from conventional FM by learning an average velocity field that directly supports accurate one-step action generation. This design removes the need for multi-step denoising and backpropagation-through-time updates, resulting in substantially faster and more robust learning. Extensive experiments on the D4RL benchmark show that OFQL, despite generating actions in a single step, not only significantly reduces computation during both training and inference but also outperforms multi-step DQL by a large margin. Furthermore, OFQL surpasses all other baselines, achieving state-of-the-art performance in D4RL.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-770444539442a1ae", "title": "Bias Testing and Mitigation in LLM-based Code Generation", "authors": ["Dong Huang", "Jie M. Zhang", "Qingwen Bu", "Xiaofei Xie", "Junjie Chen", "Heming Cui"], "year": 2023, "published": "2023-09-03T07:14:49Z", "updated": "2025-03-21T06:36:33Z", "arxiv_id": "http://arxiv.org/abs/2309.14345v4", "url": "https://arxiv.org/abs/2309.14345v4", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Deep Research"], "benchmarks": [], "models": ["GPT-4-turbo", "GPT-4", "GPT-3.5-turbo", "Claude-instant-1", "PALM-2-CodeChat-bison"], "key_results": ["As the adoption of LLMs becomes more widespread in software coding ecosystems, a pressing issue has emerged: does the generated code contain social bias and unfairness, such as those related to age, gender, and race?"], "abstract": "As the adoption of LLMs becomes more widespread in software coding ecosystems, a pressing issue has emerged: does the generated code contain social bias and unfairness, such as those related to age, gender, and race? This issue concerns the integrity, fairness, and ethical foundation of software applications that depend on the code generated by these models but are underexplored in the literature. This paper presents a novel bias testing framework that is specifically designed for code generation tasks. Based on this framework, we conduct an extensive empirical study on the biases in code generated by five widely studied LLMs (i.e., PALM-2-CodeChat-bison, Claude-instant-1, GPT-3.5-turbo, GPT-4-turbo, and GPT-4). Our findings reveal that biases are prevalent. For example, 13.47% to 49.10% of the codes generated by these LLMs have biased behaviors towards gender. Moreover, we study five bias mitigation prompt strategies that are commonly used in current code generation scenarios, i.e., zero-shot, one-shot, few-shot, and two Chain-of-Thought (CoT) prompts, with and without provided feedback-driven refinement. Our evaluation results illustrate that using direct prompt engineering strategies has limited effectiveness in mitigating bias, but our test execution feedback can help to reduce the ratio of code biases to a large extent (e.g., from 59.88% to 4.79% for GPT-4).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d5b8921bb5b8190d", "title": "Neurosymbolic Reinforcement Learning and Planning: A Survey", "authors": ["K. Acharya", "W. Raza", "C. M. J. M. Dourado", "A. Velasquez", "H. Song"], "year": 2023, "published": "2023-09-02T23:41:35Z", "updated": "2023-09-02T23:41:35Z", "arxiv_id": "http://arxiv.org/abs/2309.01038v1", "url": "https://arxiv.org/abs/2309.01038v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Deep Research"], "benchmarks": [], "models": ["ying", "yielded"], "key_results": ["The area of Neurosymbolic Artificial Intelligence (Neurosymbolic AI) is rapidly developing and has become a popular research topic, encompassing sub-fields such as Neurosymbolic Deep Learning (Neurosymbolic DL) and Neurosymbolic Reinforcement Learning (Neurosymbolic RL)."], "abstract": "The area of Neurosymbolic Artificial Intelligence (Neurosymbolic AI) is rapidly developing and has become a popular research topic, encompassing sub-fields such as Neurosymbolic Deep Learning (Neurosymbolic DL) and Neurosymbolic Reinforcement Learning (Neurosymbolic RL). Compared to traditional learning methods, Neurosymbolic AI offers significant advantages by simplifying complexity and providing transparency and explainability. Reinforcement Learning(RL), a long-standing Artificial Intelligence(AI) concept that mimics human behavior using rewards and punishment, is a fundamental component of Neurosymbolic RL, a recent integration of the two fields that has yielded promising results. The aim of this paper is to contribute to the emerging field of Neurosymbolic RL by conducting a literature survey. Our evaluation focuses on the three components that constitute Neurosymbolic RL: neural, symbolic, and RL. We categorize works based on the role played by the neural and symbolic parts in RL, into three taxonomies:Learning for Reasoning, Reasoning for Learning and Learning-Reasoning. These categories are further divided into sub-categories based on their applications. Furthermore, we analyze the RL components of each research work, including the state space, action space, policy module, and RL algorithm. Additionally, we identify research opportunities and challenges in various applications within this dynamic field.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-2f498478abdc3a5b", "title": "A Boosted Machine Learning Framework for the Improvement of Phase and Crystal Structure Prediction of High Entropy Alloys Using Thermodynamic and Configurational Parameters", "authors": ["Debsundar Dey", "Suchandan Das", "Anik Pal", "Santanu Dey", "Chandan Kumar Raul", "Arghya Chatterjee"], "year": 2023, "published": "2023-09-02T17:18:53Z", "updated": "2023-12-31T14:11:49Z", "arxiv_id": "http://arxiv.org/abs/2309.00993v2", "url": "https://arxiv.org/abs/2309.00993v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["This study employed five distinct boosting algorithms to predict phases and crystal structures, offering an enhanced guideline for improving the accuracy of these predictions.", "Among all these algorithms, XGBoost gives the highest accuracy of prediction (94.05%) for phases and LightGBM gives the highest accuracy of prediction of crystal structure of the phases (90.07%)."], "abstract": "The reason behind the remarkable properties of High-Entropy Alloys (HEAs) is rooted in the diverse phases and the crystal structures they contain. In the realm of material informatics, employing machine learning (ML) techniques to classify phases and crystal structures of HEAs has gained considerable significance. In this study, we assembled a new collection of 1345 HEAs with varying compositions to predict phases. Within this collection, there were 705 sets of data that were utilized to predict the crystal structures with the help of thermodynamics and electronic configuration. Our study introduces a methodical framework i.e., the Pearson correlation coefficient that helps in selecting the strongly co-related features to increase the prediction accuracy. This study employed five distinct boosting algorithms to predict phases and crystal structures, offering an enhanced guideline for improving the accuracy of these predictions. Among all these algorithms, XGBoost gives the highest accuracy of prediction (94.05%) for phases and LightGBM gives the highest accuracy of prediction of crystal structure of the phases (90.07%). The quantification of the influence exerted by parameters on the model's accuracy was conducted and a new approach was made to elucidate the contribution of individual parameters in the process of phase prediction and crystal structure prediction.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-3929ff5ff9ce033e", "title": "Sequential Dexterity: Chaining Dexterous Policies for Long-Horizon Manipulation", "authors": ["Yuanpei Chen", "Chen Wang", "Li Fei-Fei", "C. Karen Liu"], "year": 2023, "published": "2023-09-02T16:55:48Z", "updated": "2023-10-16T05:05:56Z", "arxiv_id": "http://arxiv.org/abs/2309.00987v2", "url": "https://arxiv.org/abs/2309.00987v2", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["We present Sequential Dexterity, a general system based on reinforcement learning (RL) that chains multiple dexterous policies for achieving long-horizon task goals."], "abstract": "Many real-world manipulation tasks consist of a series of subtasks that are significantly different from one another. Such long-horizon, complex tasks highlight the potential of dexterous hands, which possess adaptability and versatility, capable of seamlessly transitioning between different modes of functionality without the need for re-grasping or external tools. However, the challenges arise due to the high-dimensional action space of dexterous hand and complex compositional dynamics of the long-horizon tasks. We present Sequential Dexterity, a general system based on reinforcement learning (RL) that chains multiple dexterous policies for achieving long-horizon task goals. The core of the system is a transition feasibility function that progressively finetunes the sub-policies for enhancing chaining success rate, while also enables autonomous policy-switching for recovery from failures and bypassing redundant stages. Despite being trained only in simulation with a few task objects, our system demonstrates generalization capability to novel object shapes and is able to zero-shot transfer to a real-world robot equipped with a dexterous hand. Code and videos are available at https://sequential-dexterity.github.io", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bc4b1181725120e9", "title": "ModelScope-Agent: Building Your Customizable Agent System with Open-source Large Language Models", "authors": ["Chenliang Li", "Hehong Chen", "Ming Yan", "Weizhou Shen", "Haiyang Xu", "Zhikai Wu", "Zhicheng Zhang", "Wenmeng Zhou", "Yingda Chen", "Chen Cheng", "Hongzhu Shi", "Ji Zhang", "Fei Huang", "Jingren Zhou"], "year": 2023, "published": "2023-09-02T16:50:30Z", "updated": "2023-09-02T16:50:30Z", "arxiv_id": "http://arxiv.org/abs/2309.00986v1", "url": "https://arxiv.org/abs/2309.00986v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Tool Use", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Large language models (LLMs) have recently demonstrated remarkable capabilities to comprehend human intentions, engage in reasoning, and design planning-like behavior."], "abstract": "Large language models (LLMs) have recently demonstrated remarkable capabilities to comprehend human intentions, engage in reasoning, and design planning-like behavior. To further unleash the power of LLMs to accomplish complex tasks, there is a growing trend to build agent framework that equips LLMs, such as ChatGPT, with tool-use abilities to connect with massive external APIs. In this work, we introduce ModelScope-Agent, a general and customizable agent framework for real-world applications, based on open-source LLMs as controllers. It provides a user-friendly system library, with customizable engine design to support model training on multiple open-source LLMs, while also enabling seamless integration with both model APIs and common APIs in a unified way. To equip the LLMs with tool-use abilities, a comprehensive framework has been proposed spanning over tool-use data collection, tool retrieval, tool registration, memory control, customized model training, and evaluation for practical real-world applications. Finally, we showcase ModelScopeGPT, a real-world intelligent assistant of ModelScope Community based on the ModelScope-Agent framework, which is able to connect open-source LLMs with more than 1000 public AI models and localized community knowledge in ModelScope. The ModelScope-Agent library\\footnote{https://github.com/modelscope/modelscope-agent} and online demo\\footnote{https://modelscope.cn/studios/damo/ModelScopeGPT/summary} are now publicly available.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0401edf59eb87cc3", "title": "Compositional Diffusion-Based Continuous Constraint Solvers", "authors": ["Zhutian Yang", "Jiayuan Mao", "Yilun Du", "Jiajun Wu", "Joshua B. Tenenbaum", "Tomás Lozano-Pérez", "Leslie Pack Kaelbling"], "year": 2023, "published": "2023-09-02T15:20:36Z", "updated": "2023-09-02T15:20:36Z", "arxiv_id": "http://arxiv.org/abs/2309.00966v1", "url": "https://arxiv.org/abs/2309.00966v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["This paper introduces an approach for learning to solve continuous constraint satisfaction problems (CCSP) in robotic reasoning and planning."], "abstract": "This paper introduces an approach for learning to solve continuous constraint satisfaction problems (CCSP) in robotic reasoning and planning. Previous methods primarily rely on hand-engineering or learning generators for specific constraint types and then rejecting the value assignments when other constraints are violated. By contrast, our model, the compositional diffusion continuous constraint solver (Diffusion-CCSP) derives global solutions to CCSPs by representing them as factor graphs and combining the energies of diffusion models trained to sample for individual constraint types. Diffusion-CCSP exhibits strong generalization to novel combinations of known constraints, and it can be integrated into a task and motion planner to devise long-horizon plans that include actions with both discrete and continuous parameters. Project site: https://diffusion-ccsp.github.io/", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-07e27cd4776bf0ea", "title": "Multilingual Text Representation", "authors": ["Fahim Faisal"], "year": 2023, "published": "2023-09-02T14:21:22Z", "updated": "2023-09-02T14:21:22Z", "arxiv_id": "http://arxiv.org/abs/2309.00949v1", "url": "https://arxiv.org/abs/2309.00949v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["State-of-the-art language models came a long way, starting from the simple one-hot representation of words capable of performing tasks like natural language understanding, common-sense reasoning, or question-answering, thus capturing both the syntax and semantics of texts.", "In this survey, we shed light on this iterative progression of multilingual text representation and discuss the driving factors that ultimately led to the current state-of-the-art.", "Subsequently, we discuss how the full potential of language democratization could be obtained, reaching beyond the known limits and what is the scope of improvement in that space."], "abstract": "Modern NLP breakthrough includes large multilingual models capable of performing tasks across more than 100 languages. State-of-the-art language models came a long way, starting from the simple one-hot representation of words capable of performing tasks like natural language understanding, common-sense reasoning, or question-answering, thus capturing both the syntax and semantics of texts. At the same time, language models are expanding beyond our known language boundary, even competitively performing over very low-resource dialects of endangered languages. However, there are still problems to solve to ensure an equitable representation of texts through a unified modeling space across language and speakers. In this survey, we shed light on this iterative progression of multilingual text representation and discuss the driving factors that ultimately led to the current state-of-the-art. Subsequently, we discuss how the full potential of language democratization could be obtained, reaching beyond the known limits and what is the scope of improvement in that space.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-70a5625b49f1eff2", "title": "Large Process Models: A Vision for Business Process Management in the Age of Generative AI", "authors": ["Timotheus Kampik", "Christian Warmuth", "Adrian Rebmann", "Ron Agam", "Lukas N. P. Egger", "Andreas Gerber", "Johannes Hoffart", "Jonas Kolk", "Philipp Herzig", "Gero Decker", "Han van der Aa", "Artem Polyvyanyy", "Stefanie Rinderle-Ma", "Ingo Weber", "Matthias Weidlich"], "year": 2023, "published": "2023-09-02T10:32:53Z", "updated": "2025-01-17T11:18:37Z", "arxiv_id": "http://arxiv.org/abs/2309.00900v3", "url": "https://arxiv.org/abs/2309.00900v3", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["In this vision, the proposed LPM would allow organizations to receive context-specific (tailored) process and other business models, analytical deep-dives, and improvement recommendations."], "abstract": "The continued success of Large Language Models (LLMs) and other generative artificial intelligence approaches highlights the advantages that large information corpora can have over rigidly defined symbolic models, but also serves as a proof-point of the challenges that purely statistics-based approaches have in terms of safety and trustworthiness. As a framework for contextualizing the potential, as well as the limitations of LLMs and other foundation model-based technologies, we propose the concept of a Large Process Model (LPM) that combines the correlation power of LLMs with the analytical precision and reliability of knowledge-based systems and automated reasoning approaches. LPMs are envisioned to directly utilize the wealth of process management experience that experts have accumulated, as well as process performance data of organizations with diverse characteristics, e.g.,\\ regarding size, region, or industry. In this vision, the proposed LPM would allow organizations to receive context-specific (tailored) process and other business models, analytical deep-dives, and improvement recommendations. As such, they would allow to substantially decrease the time and effort required for business transformation, while also allowing for deeper, more impactful, and more actionable insights than previously possible. We argue that implementing an LPM is feasible, but also highlight limitations and research challenges that need to be solved to implement particular aspects of the LPM vision.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-21c7cd108fd45a9c", "title": "A Generic Fundus Image Enhancement Network Boosted by Frequency Self-supervised Representation Learning", "authors": ["Heng Li", "Haofeng Liu", "Huazhu Fu", "Yanwu Xu", "Hui Shu", "Ke Niu", "Yan Hu", "Jiang Liu"], "year": 2023, "published": "2023-09-02T09:51:30Z", "updated": "2023-09-02T09:51:30Z", "arxiv_id": "http://arxiv.org/abs/2309.00885v1", "url": "https://arxiv.org/abs/2309.00885v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["Compared with state-of-the-art algorithms, GFE-Net achieves superior performance in data dependency, enhancement performance, deployment efficiency, and scale generalizability."], "abstract": "Fundus photography is prone to suffer from image quality degradation that impacts clinical examination performed by ophthalmologists or intelligent systems. Though enhancement algorithms have been developed to promote fundus observation on degraded images, high data demands and limited applicability hinder their clinical deployment. To circumvent this bottleneck, a generic fundus image enhancement network (GFE-Net) is developed in this study to robustly correct unknown fundus images without supervised or extra data. Levering image frequency information, self-supervised representation learning is conducted to learn robust structure-aware representations from degraded images. Then with a seamless architecture that couples representation learning and image enhancement, GFE-Net can accurately correct fundus images and meanwhile preserve retinal structures. Comprehensive experiments are implemented to demonstrate the effectiveness and advantages of GFE-Net. Compared with state-of-the-art algorithms, GFE-Net achieves superior performance in data dependency, enhancement performance, deployment efficiency, and scale generalizability. Follow-up fundus image analysis is also facilitated by GFE-Net, whose modules are respectively verified to be effective for image enhancement.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-39c71b758ce0ee49", "title": "Towards Certified Probabilistic Robustness with High Accuracy", "authors": ["Ruihan Zhang", "Peixin Zhang", "Jun Sun"], "year": 2023, "published": "2023-09-02T09:39:47Z", "updated": "2023-09-02T09:39:47Z", "arxiv_id": "http://arxiv.org/abs/2309.00879v1", "url": "https://arxiv.org/abs/2309.00879v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": ["DROP"], "models": [], "key_results": ["For example, adversarial training improves empirical robustness, but they do not provide certification of the model's robustness.", "In this work, we propose a novel approach that aims to achieve both high accuracy and certified probabilistic robustness.", "Our experiments on multiple models trained on different datasets demonstrate that our approach significantly outperforms existing approaches in terms of both certification rate and accuracy."], "abstract": "Adversarial examples pose a security threat to many critical systems built on neural networks (such as face recognition systems, and self-driving cars). While many methods have been proposed to build robust models, how to build certifiably robust yet accurate neural network models remains an open problem. For example, adversarial training improves empirical robustness, but they do not provide certification of the model's robustness. On the other hand, certified training provides certified robustness but at the cost of a significant accuracy drop. In this work, we propose a novel approach that aims to achieve both high accuracy and certified probabilistic robustness. Our method has two parts, i.e., a probabilistic robust training method with an additional goal of minimizing variance in terms of divergence and a runtime inference method for certified probabilistic robustness of the prediction. The latter enables efficient certification of the model's probabilistic robustness at runtime with statistical guarantees. This is supported by our training objective, which minimizes the variance of the model's predictions in a given vicinity, derived from a general definition of model robustness. Our approach works for a variety of perturbations and is reasonably efficient. Our experiments on multiple models trained on different datasets demonstrate that our approach significantly outperforms existing approaches in terms of both certification rate and accuracy.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-26e4a76b01ba509d", "title": "pSTarC: Pseudo Source Guided Target Clustering for Fully Test-Time Adaptation", "authors": ["Manogna Sreenivas", "Goirik Chakrabarty", "Soma Biswas"], "year": 2023, "published": "2023-09-02T07:13:47Z", "updated": "2023-11-22T11:02:35Z", "arxiv_id": "http://arxiv.org/abs/2309.00846v2", "url": "https://arxiv.org/abs/2309.00846v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Test-time Scaling"], "benchmarks": [], "models": [], "key_results": ["This method exhibits significant improvements in prediction accuracy along with efficient computational requirements."], "abstract": "Test Time Adaptation (TTA) is a pivotal concept in machine learning, enabling models to perform well in real-world scenarios, where test data distribution differs from training. In this work, we propose a novel approach called pseudo Source guided Target Clustering (pSTarC) addressing the relatively unexplored area of TTA under real-world domain shifts. This method draws inspiration from target clustering techniques and exploits the source classifier for generating pseudo-source samples. The test samples are strategically aligned with these pseudo-source samples, facilitating their clustering and thereby enhancing TTA performance. pSTarC operates solely within the fully test-time adaptation protocol, removing the need for actual source data. Experimental validation on a variety of domain shift datasets, namely VisDA, Office-Home, DomainNet-126, CIFAR-100C verifies pSTarC's effectiveness. This method exhibits significant improvements in prediction accuracy along with efficient computational requirements. Furthermore, we also demonstrate the universality of the pSTarC framework by showing its effectiveness for the continuous TTA framework. The source code for our method is available at https://manogna-s.github.io/pstarc", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e6767168a9250a3b", "title": "LeanContext: Cost-Efficient Domain-Specific Question Answering Using LLMs", "authors": ["Md Adnan Arefeen", "Biplob Debnath", "Srimat Chakradhar"], "year": 2023, "published": "2023-09-02T06:33:18Z", "updated": "2023-09-02T06:33:18Z", "arxiv_id": "http://arxiv.org/abs/2309.00841v1", "url": "https://arxiv.org/abs/2309.00841v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.IR"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["We evaluate LeanContext against several recent query-aware and query-unaware context reduction approaches on prominent datasets (arxiv papers and BBC news articles)."], "abstract": "Question-answering (QA) is a significant application of Large Language Models (LLMs), shaping chatbot capabilities across healthcare, education, and customer service. However, widespread LLM integration presents a challenge for small businesses due to the high expenses of LLM API usage. Costs rise rapidly when domain-specific data (context) is used alongside queries for accurate domain-specific LLM responses. One option is to summarize the context by using LLMs and reduce the context. However, this can also filter out useful information that is necessary to answer some domain-specific queries. In this paper, we shift from human-oriented summarizers to AI model-friendly summaries. Our approach, LeanContext, efficiently extracts $k$ key sentences from the context that are closely aligned with the query. The choice of $k$ is neither static nor random; we introduce a reinforcement learning technique that dynamically determines $k$ based on the query and context. The rest of the less important sentences are reduced using a free open source text reduction method. We evaluate LeanContext against several recent query-aware and query-unaware context reduction approaches on prominent datasets (arxiv papers and BBC news articles). Despite cost reductions of $37.29\\%$ to $67.81\\%$, LeanContext's ROUGE-1 score decreases only by $1.41\\%$ to $2.65\\%$ compared to a baseline that retains the entire context (no summarization). Additionally, if free pretrained LLM-based summarizers are used to reduce context (into human consumable summaries), LeanContext can further modify the reduced context to enhance the accuracy (ROUGE-1 score) by $13.22\\%$ to $24.61\\%$.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-cb29bd0abdc1cccd", "title": "Multi-agent Collective Construction using 3D Decomposition", "authors": ["Akshaya Kesarimangalam Srinivasan", "Shambhavi Singh", "Geordan Gutow", "Howie Choset", "Bhaskar Vundurthy"], "year": 2023, "published": "2023-09-02T16:49:17Z", "updated": "2023-09-02T16:49:17Z", "arxiv_id": "http://arxiv.org/abs/2309.00985v1", "url": "https://arxiv.org/abs/2309.00985v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Multi-Agent", "Deep Research"], "benchmarks": ["DROP"], "models": ["ying"], "key_results": ["Extensive testing on 200 randomly generated structures shows an order of magnitude improvement in the solution computation time compared to an MILP approach without decomposition.", "Additionally, compared to Reinforcement Learning (RL) based and heuristics-based approaches drawn from the literature, our solution indicates orders of magnitude improvement in the number of pick-up and drop-off actions required to construct a structure.", "With this parallelization technique, we illustrate a further improvement in the number of time steps required to complete building the structure."], "abstract": "This paper addresses a Multi-Agent Collective Construction (MACC) problem that aims to build a three-dimensional structure comprised of cubic blocks. We use cube-shaped robots that can carry one cubic block at a time, and move forward, reverse, left, and right to an adjacent cell of the same height or climb up and down one cube height. To construct structures taller than one cube, the robots must build supporting stairs made of blocks and remove the stairs once the structure is built. Conventional techniques solve for the entire structure at once and quickly become intractable for larger workspaces and complex structures, especially in a multi-agent setting. To this end, we present a decomposition algorithm that computes valid substructures based on intrinsic structural dependencies. We use Mixed Integer Linear Programming (MILP) to solve for each of these substructures and then aggregate the solutions to construct the entire structure. Extensive testing on 200 randomly generated structures shows an order of magnitude improvement in the solution computation time compared to an MILP approach without decomposition. Additionally, compared to Reinforcement Learning (RL) based and heuristics-based approaches drawn from the literature, our solution indicates orders of magnitude improvement in the number of pick-up and drop-off actions required to construct a structure. Furthermore, we leverage the independence between substructures to detect which sub-structures can be built in parallel. With this parallelization technique, we illustrate a further improvement in the number of time steps required to complete building the structure. This work is a step towards applying multi-agent collective construction for real-world structures by significantly reducing solution computation time with a bounded increase in the number of time steps required to build the structure.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-215c1b0d240cfbd4", "title": "Trustworthy Distributed Average Consensus based on Locally Assessed Trust Evaluations", "authors": ["Christoforos N. Hadjicostis", "Alejandro D. Dominguez-Garcia"], "year": 2023, "published": "2023-09-02T11:55:30Z", "updated": "2023-09-02T11:55:30Z", "arxiv_id": "http://arxiv.org/abs/2309.00920v1", "url": "https://arxiv.org/abs/2309.00920v1", "primary_category": "cs.MA", "categories": ["cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["This paper proposes a distributed algorithm for average consensus in a multi-agent system under a fixed bidirectional communication topology, in the presence of malicious agents (nodes) that may try to influence the average consensus outcome by manipulating their updates."], "abstract": "This paper proposes a distributed algorithm for average consensus in a multi-agent system under a fixed bidirectional communication topology, in the presence of malicious agents (nodes) that may try to influence the average consensus outcome by manipulating their updates. The proposed algorithm converges asymptotically to the average of the initial values of the non-malicious nodes, which we refer to as the trustworthy average, as long as the underlying topology that describes the information exchange among the non-malicious nodes is connected. We first present a distributed iterative algorithm that assumes that each node receives (at each iteration or periodically) side information about the trustworthiness of the other nodes, and it uses such trust assessments to determine whether or not to incorporate messages received from its neighbors, as well as to make proper adjustments in its calculation depending on whether a previously trustworthy neighbor becomes untrustworthy or vice-versa. We show that, as long as the trust assessments for each non-malicious node eventually reflect correctly the status (malicious or non-malicious) of its neighboring nodes, the algorithm guarantees asymptotic convergence to the trustworthy average. We subsequently discuss how the proposed algorithm can be enhanced with functionality that enables each node to obtain trust assessments about its neighbors by utilizing information that it receives from its two-hop neighbors at infrequent, perhaps randomly chosen, time instants.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b9ec98ccbb06bc1f", "title": "Robust Online Learning over Networks", "authors": ["Nicola Bastianello", "Diego Deplano", "Mauro Franceschelli", "Karl H. Johansson"], "year": 2023, "published": "2023-09-01T15:18:05Z", "updated": "2024-05-17T11:25:19Z", "arxiv_id": "http://arxiv.org/abs/2309.00520v2", "url": "https://arxiv.org/abs/2309.00520v2", "primary_category": "math.OC", "categories": ["math.OC", "cs.LG", "cs.MA", "eess.SY"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["We corroborate the theoretical analysis with numerical simulations comparing DOT-ADMM with other state-of-the-art algorithms, showing that only the proposed algorithm exhibits robustness to (i)-(iv)."], "abstract": "The recent deployment of multi-agent networks has enabled the distributed solution of learning problems, where agents cooperate to train a global model without sharing their local, private data. This work specifically targets some prevalent challenges inherent to distributed learning: (i) online training, i.e., the local data change over time; (ii) asynchronous agent computations; (iii) unreliable and limited communications; and (iv) inexact local computations. To tackle these challenges, we apply the Distributed Operator Theoretical (DOT) version of the Alternating Direction Method of Multipliers (ADMM), which we call \"DOT-ADMM\". We prove that if the DOT-ADMM operator is metric subregular, then it converges with a linear rate for a large class of (not necessarily strongly) convex learning problems toward a bounded neighborhood of the optimal time-varying solution, and characterize how such neighborhood depends on (i)-(iv). We first derive an easy-to-verify condition for ensuring the metric subregularity of an operator, followed by tutorial examples on linear and logistic regression problems. We corroborate the theoretical analysis with numerical simulations comparing DOT-ADMM with other state-of-the-art algorithms, showing that only the proposed algorithm exhibits robustness to (i)-(iv).", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-da77217be12a22a6", "title": "Discrete Versus Continuous Algorithms in Dynamics of Affective Decision Making", "authors": ["V. I. Yukalov", "E. P. Yukalova"], "year": 2023, "published": "2023-09-01T09:23:02Z", "updated": "2023-09-01T09:23:02Z", "arxiv_id": "http://arxiv.org/abs/2309.00357v1", "url": "https://arxiv.org/abs/2309.00357v1", "primary_category": "cs.AI", "categories": ["cs.AI", "physics.soc-ph"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["The dynamics of affective decision making is considered for an intelligent network composed of agents with different types of memory: long-term and short-term memory."], "abstract": "The dynamics of affective decision making is considered for an intelligent network composed of agents with different types of memory: long-term and short-term memory. The consideration is based on probabilistic affective decision theory, which takes into account the rational utility of alternatives as well as the emotional alternative attractiveness. The objective of this paper is the comparison of two multistep operational algorithms of the intelligent network: one based on discrete dynamics and the other on continuous dynamics. By means of numerical analysis, it is shown that, depending on the network parameters, the characteristic probabilities for continuous and discrete operations can exhibit either close or drastically different behavior. Thus, depending on which algorithm is employed, either discrete or continuous, theoretical predictions can be rather different, which does not allow for a uniquely defined description of practical problems. This finding is important for understanding which of the algorithms is more appropriate for the correct analysis of decision-making tasks. A discussion is given, revealing that the discrete operation seems to be more realistic for describing intelligent networks as well as affective artificial intelligence.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e3ae1a465af8f056", "title": "Implementing BDI Continual Temporal Planning for Robotic Agents", "authors": ["Alex Zanetti", "Devis Dal Moro", "Redi Vreto", "Marco Robol", "Marco Roveri", "Paolo Giorgini"], "year": 2023, "published": "2023-09-01T08:27:44Z", "updated": "2023-09-01T08:27:44Z", "arxiv_id": "http://arxiv.org/abs/2309.00327v1", "url": "https://arxiv.org/abs/2309.00327v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Deep Research"], "benchmarks": [], "models": [], "key_results": ["Reacting to events is still a fundamental ability for an agent, but it has to be boosted up with proactive behaviors that allow the agent to explore alternatives and decide at run-time for optimal solutions."], "abstract": "Making autonomous agents effective in real-life applications requires the ability to decide at run-time and a high degree of adaptability to unpredictable and uncontrollable events. Reacting to events is still a fundamental ability for an agent, but it has to be boosted up with proactive behaviors that allow the agent to explore alternatives and decide at run-time for optimal solutions. This calls for a continuous planning as part of the deliberation process that makes an agent able to reconsider plans on the base of temporal constraints and changes of the environment. Online planning literature offers several approaches used to select the next action on the base of a partial exploration of the solution space. In this paper, we propose a BDI continuous temporal planning framework, where interleave planning and execution loop is used to integrate online planning with the BDI control-loop. The framework has been implemented with the ROS2 robotic framework and planning algorithms offered by JavaFF.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9e4d33562515fa6a", "title": "Sherlock Holmes Doesn't Play Dice: The mathematics of uncertain reasoning when something may happen, that one is not even able to figure out", "authors": ["Guido Fioretti"], "year": 2023, "published": "2023-09-01T07:52:28Z", "updated": "2025-03-28T11:11:28Z", "arxiv_id": "http://arxiv.org/abs/2309.03222v3", "url": "https://arxiv.org/abs/2309.03222v3", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.HC"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["phisticated"], "key_results": ["While Evidence Theory (also known as Dempster-Shafer Theory, or Belief Functions Theory) is being increasingly used in data fusion, its potentialities in the Social and Life Sciences are often obscured by lack of awareness of its distinctive features."], "abstract": "While Evidence Theory (also known as Dempster-Shafer Theory, or Belief Functions Theory) is being increasingly used in data fusion, its potentialities in the Social and Life Sciences are often obscured by lack of awareness of its distinctive features. In particular, with this paper I stress that an extended version of Evidence Theory can express the uncertainty deriving from the fear that events may materialize, that one is not even able to figure out. By contrast, Probability Theory must limit itself to the possibilities that a decision-maker is currently envisaging.\n  I compare this extended version of Evidence Theory to sophisticated extensions of Probability Theory, such as imprecise and sub-additive probabilities, as well as unconventional versions of Information Theory that are employed in data fusion and transmission of cultural information. A further extension to multi-agent interaction is outlined.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-93840a08138bc4fc", "title": "Convergence Analysis of the Best Response Algorithm for Time-Varying Games", "authors": ["Zifan Wang", "Yi Shen", "Michael M. Zavlanos", "Karl H. Johansson"], "year": 2023, "published": "2023-09-01T07:35:10Z", "updated": "2023-09-01T07:35:10Z", "arxiv_id": "http://arxiv.org/abs/2309.00307v1", "url": "https://arxiv.org/abs/2309.00307v1", "primary_category": "math.OC", "categories": ["math.OC", "cs.GT", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["Specifically, we provide a sufficient condition on the strong monotonicity parameter of the time-invariant games under which the proposed best response algorithm achieves exponential convergence to the static Nash equilibrium."], "abstract": "This paper studies a class of strongly monotone games involving non-cooperative agents that optimize their own time-varying cost functions. We assume that the agents can observe other agents' historical actions and choose actions that best respond to other agents' previous actions; we call this a best response scheme. We start by analyzing the convergence rate of this best response scheme for standard time-invariant games. Specifically, we provide a sufficient condition on the strong monotonicity parameter of the time-invariant games under which the proposed best response algorithm achieves exponential convergence to the static Nash equilibrium. We further illustrate that this best response algorithm may oscillate when the proposed sufficient condition fails to hold, which indicates that this condition is tight. Next, we analyze this best response algorithm for time-varying games where the cost functions of each agent change over time. Under similar conditions as for time-invariant games, we show that the proposed best response algorithm stays asymptotically close to the evolving equilibrium. We do so by analyzing both the equilibrium tracking error and the dynamic regret. Numerical experiments on economic market problems are presented to validate our analysis.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ff1a2342d0f07c34", "title": "End-to-end Lidar-Driven Reinforcement Learning for Autonomous Racing", "authors": ["Meraj Mammadov"], "year": 2023, "published": "2023-09-01T07:03:05Z", "updated": "2023-09-01T07:03:05Z", "arxiv_id": "http://arxiv.org/abs/2309.00296v1", "url": "https://arxiv.org/abs/2309.00296v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["phisticated"], "key_results": ["Reinforcement Learning (RL) has emerged as a transformative approach in the domains of automation and robotics, offering powerful solutions to complex problems that conventional methods struggle to address."], "abstract": "Reinforcement Learning (RL) has emerged as a transformative approach in the domains of automation and robotics, offering powerful solutions to complex problems that conventional methods struggle to address. In scenarios where the problem definitions are elusive and challenging to quantify, learning-based solutions such as RL become particularly valuable. One instance of such complexity can be found in the realm of car racing, a dynamic and unpredictable environment that demands sophisticated decision-making algorithms. This study focuses on developing and training an RL agent to navigate a racing environment solely using feedforward raw lidar and velocity data in a simulated context. The agent's performance, trained in the simulation environment, is then experimentally evaluated in a real-world racing scenario. This exploration underlines the feasibility and potential benefits of RL algorithm enhancing autonomous racing performance, especially in the environments where prior map information is not available.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-a36f338e8b149082", "title": "Detecting Evidence of Organization in groups by Trajectories", "authors": ["T. F. Silva", "J. E. B. Maia"], "year": 2023, "published": "2023-08-31T23:57:02Z", "updated": "2023-08-31T23:57:02Z", "arxiv_id": "http://arxiv.org/abs/2309.00172v1", "url": "https://arxiv.org/abs/2309.00172v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Safety / Jailbreak", "Deep Research"], "benchmarks": [], "models": ["ying"], "key_results": ["Effective detection of organizations is essential for fighting crime and maintaining public safety, especially considering the limited human resources and tools to deal with each group that exhibits co-movement patterns."], "abstract": "Effective detection of organizations is essential for fighting crime and maintaining public safety, especially considering the limited human resources and tools to deal with each group that exhibits co-movement patterns. This paper focuses on solving the Network Structure Inference (NSI) challenge. Thus, we introduce two new approaches to detect network structure inferences based on agent trajectories. The first approach is based on the evaluation of graph entropy, while the second considers the quality of clustering indices. To evaluate the effectiveness of the new approaches, we conducted experiments using four scenario simulations based on the animal kingdom, available on the NetLogo platform: Ants, Wolf Sheep Predation, Flocking, and Ant Adaptation. Furthermore, we compare the results obtained with those of an approach previously proposed in the literature, applying all methods to simulations of the NetLogo platform. The results demonstrate that our new detection approaches can more clearly identify the inferences of organizations or networks in the simulated scenarios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b735826fe12ef7f8", "title": "Construction Grammar and Artificial Intelligence", "authors": ["Katrien Beuls", "Paul Van Eecke"], "year": 2023, "published": "2023-08-31T21:15:06Z", "updated": "2023-08-31T21:15:06Z", "arxiv_id": "http://arxiv.org/abs/2309.00135v1", "url": "https://arxiv.org/abs/2309.00135v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["In this chapter, we argue that it is highly beneficial for the contemporary construction grammarian to have a thorough understanding of the strong relationship between the research fields of construction grammar and artificial intelligence."], "abstract": "In this chapter, we argue that it is highly beneficial for the contemporary construction grammarian to have a thorough understanding of the strong relationship between the research fields of construction grammar and artificial intelligence. We start by unravelling the historical links between the two fields, showing that their relationship is rooted in a common attitude towards human communication and language. We then discuss the first direction of influence, focussing in particular on how insights and techniques from the field of artificial intelligence play an important role in operationalising, validating and scaling constructionist approaches to language. We then proceed to the second direction of influence, highlighting the relevance of construction grammar insights and analyses to the artificial intelligence endeavour of building truly intelligent agents. We support our case with a variety of illustrative examples and conclude that the further elaboration of this relationship will play a key role in shaping the future of the field of construction grammar.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-4ff3bdf4a6040661", "title": "Agent Teaming Situation Awareness (ATSA): A Situation Awareness Framework for Human-AI Teaming", "authors": ["Qi Gao", "Wei Xu", "Mowei Shen", "Zaifeng Gao"], "year": 2023, "published": "2023-08-31T15:02:01Z", "updated": "2023-09-04T12:23:36Z", "arxiv_id": "http://arxiv.org/abs/2308.16785v2", "url": "https://arxiv.org/abs/2308.16785v2", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.HC"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["The rapid advancements in artificial intelligence (AI) have led to a growing trend of human-AI teaming (HAT) in various fields."], "abstract": "The rapid advancements in artificial intelligence (AI) have led to a growing trend of human-AI teaming (HAT) in various fields. As machines continue to evolve from mere automation to a state of autonomy, they are increasingly exhibiting unexpected behaviors and human-like cognitive/intelligent capabilities, including situation awareness (SA). This shift has the potential to enhance the performance of mixed human-AI teams over all-human teams, underscoring the need for a better understanding of the dynamic SA interactions between humans and machines. To this end, we provide a review of leading SA theoretical models and a new framework for SA in the HAT context based on the key features and processes of HAT. The Agent Teaming Situation Awareness (ATSA) framework unifies human and AI behavior, and involves bidirectional, and dynamic interaction. The framework is based on the individual and team SA models and elaborates on the cognitive mechanisms for modeling HAT. Similar perceptual cycles are adopted for the individual (including both human and AI) and the whole team, which is tailored to the unique requirements of the HAT context. ATSA emphasizes cohesive and effective HAT through structures and components, including teaming understanding, teaming control, and the world, as well as adhesive transactive part. We further propose several future research directions to expand on the distinctive contributions of ATSA and address the specific and pressing next steps.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5e28260395a7abf3", "title": "On a Connection between Differential Games, Optimal Control, and Energy-based Models for Multi-Agent Interactions", "authors": ["Christopher Diehl", "Tobias Klosek", "Martin Krüger", "Nils Murzyn", "Torsten Bertram"], "year": 2023, "published": "2023-08-31T08:30:11Z", "updated": "2023-10-16T14:37:54Z", "arxiv_id": "http://arxiv.org/abs/2308.16539v2", "url": "https://arxiv.org/abs/2308.16539v2", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.GT", "cs.LG", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Interpretability", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["The experiments using simulated mobile robot pedestrian interactions and real-world automated driving data provide empirical evidence that the game-theoretic layer improves the predictive performance of various neural network backbones."], "abstract": "Game theory offers an interpretable mathematical framework for modeling multi-agent interactions. However, its applicability in real-world robotics applications is hindered by several challenges, such as unknown agents' preferences and goals. To address these challenges, we show a connection between differential games, optimal control, and energy-based models and demonstrate how existing approaches can be unified under our proposed Energy-based Potential Game formulation. Building upon this formulation, this work introduces a new end-to-end learning application that combines neural networks for game-parameter inference with a differentiable game-theoretic optimization layer, acting as an inductive bias. The experiments using simulated mobile robot pedestrian interactions and real-world automated driving data provide empirical evidence that the game-theoretic layer improves the predictive performance of various neural network backbones.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-33a089db9a420e7b", "title": "Recommender AI Agent: Integrating Large Language Models for Interactive Recommendations", "authors": ["Xu Huang", "Jianxun Lian", "Yuxuan Lei", "Jing Yao", "Defu Lian", "Xing Xie"], "year": 2023, "published": "2023-08-31T07:36:44Z", "updated": "2024-01-30T03:17:26Z", "arxiv_id": "http://arxiv.org/abs/2308.16505v3", "url": "https://arxiv.org/abs/2308.16505v3", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG"], "benchmarks": [], "models": ["ying"], "key_results": ["Experimental results on several public datasets show that InteRecAgent achieves satisfying performance as a conversational recommender system, outperforming general-purpose LLMs."], "abstract": "Recommender models excel at providing domain-specific item recommendations by leveraging extensive user behavior data. Despite their ability to act as lightweight domain experts, they struggle to perform versatile tasks such as providing explanations and engaging in conversations. On the other hand, large language models (LLMs) represent a significant step towards artificial general intelligence, showcasing remarkable capabilities in instruction comprehension, commonsense reasoning, and human interaction. However, LLMs lack the knowledge of domain-specific item catalogs and behavioral patterns, particularly in areas that diverge from general world knowledge, such as online e-commerce. Finetuning LLMs for each domain is neither economic nor efficient.\n  In this paper, we bridge the gap between recommender models and LLMs, combining their respective strengths to create a versatile and interactive recommender system. We introduce an efficient framework called \\textbf{InteRecAgent}, which employs LLMs as the brain and recommender models as tools. We first outline a minimal set of essential tools required to transform LLMs into InteRecAgent. We then propose an efficient workflow within InteRecAgent for task execution, incorporating key components such as memory components, dynamic demonstration-augmented task planning, and reflection. InteRecAgent enables traditional recommender systems, such as those ID-based matrix factorization models, to become interactive systems with a natural language interface through the integration of LLMs. Experimental results on several public datasets show that InteRecAgent achieves satisfying performance as a conversational recommender system, outperforming general-purpose LLMs. The source code of InteRecAgent is released at https://aka.ms/recagent.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-f72c117d41511757", "title": "Individually Rational Collaborative Vehicle Routing through Give-And-Take Exchanges", "authors": ["Paul Mingzheng Tang", "Ba Phong Tran", "Hoong Chuin Lau"], "year": 2023, "published": "2023-08-31T07:18:37Z", "updated": "2023-08-31T07:18:37Z", "arxiv_id": "http://arxiv.org/abs/2308.16501v1", "url": "https://arxiv.org/abs/2308.16501v1", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["In this paper, we are concerned with the automated exchange of orders between logistics companies in a marketplace platform to optimize total revenues."], "abstract": "In this paper, we are concerned with the automated exchange of orders between logistics companies in a marketplace platform to optimize total revenues. We introduce a novel multi-agent approach to this problem, focusing on the Collaborative Vehicle Routing Problem (CVRP) through the lens of individual rationality. Our proposed algorithm applies the principles of Vehicle Routing Problem (VRP) to pairs of vehicles from different logistics companies, optimizing the overall routes while considering standard VRP constraints plus individual rationality constraints. By facilitating cooperation among competing logistics agents through a Give-and-Take approach, we show that it is possible to reduce travel distance and increase operational efficiency system-wide. More importantly, our approach ensures individual rationality and faster convergence, which are important properties of ensuring the long-term sustainability of the marketplace platform. We demonstrate the efficacy of our approach through extensive experiments using real-world test data from major logistics companies. The results reveal our algorithm's ability to rapidly identify numerous optimal solutions, underscoring its practical applicability and potential to transform the logistics industry.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b4c94111be7452a6", "title": "Causal Strategic Learning with Competitive Selection", "authors": ["Kiet Q. H. Vo", "Muneeb Aadil", "Siu Lun Chau", "Krikamol Muandet"], "year": 2023, "published": "2023-08-30T18:43:11Z", "updated": "2024-02-03T22:44:45Z", "arxiv_id": "http://arxiv.org/abs/2308.16262v3", "url": "https://arxiv.org/abs/2308.16262v3", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["When each decision maker unilaterally selects agents by maximising their own utility, we show that the optimal selection rule is a trade-off between selecting the best agents and providing incentives to maximise the agents' improvement."], "abstract": "We study the problem of agent selection in causal strategic learning under multiple decision makers and address two key challenges that come with it. Firstly, while much of prior work focuses on studying a fixed pool of agents that remains static regardless of their evaluations, we consider the impact of selection procedure by which agents are not only evaluated, but also selected. When each decision maker unilaterally selects agents by maximising their own utility, we show that the optimal selection rule is a trade-off between selecting the best agents and providing incentives to maximise the agents' improvement. Furthermore, this optimal selection rule relies on incorrect predictions of agents' outcomes. Hence, we study the conditions under which a decision maker's optimal selection rule will not lead to deterioration of agents' outcome nor cause unjust reduction in agents' selection chance. To that end, we provide an analytical form of the optimal selection rule and a mechanism to retrieve the causal parameters from observational data, under certain assumptions on agents' behaviour. Secondly, when there are multiple decision makers, the interference between selection rules introduces another source of biases in estimating the underlying causal parameters. To address this problem, we provide a cooperative protocol which all decision makers must collectively adopt to recover the true causal parameters. Lastly, we complement our theoretical results with simulation studies. Our results highlight not only the importance of causal modeling as a strategy to mitigate the effect of gaming, as suggested by previous work, but also the need of a benevolent regulator to enable it.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-799971d990bb4222", "title": "EnsembleFollower: A Hybrid Car-Following Framework Based On Reinforcement Learning and Hierarchical Planning", "authors": ["Xu Han", "Xianda Chen", "Meixin Zhu", "Pinlong Cai", "Jianshan Zhou", "Xiaowen Chu"], "year": 2023, "published": "2023-08-30T12:55:02Z", "updated": "2023-08-30T12:55:02Z", "arxiv_id": "http://arxiv.org/abs/2308.16008v1", "url": "https://arxiv.org/abs/2308.16008v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["yields"], "key_results": ["Therefore, we propose EnsembleFollower, a hierarchical planning framework for achieving advanced human-like car-following.", "The experimental results illustrate that EnsembleFollower yields improved accuracy of human-like behavior and achieves effectiveness in combining hybrid models, demonstrating that our proposed framework can handle diverse car-following conditions by leveraging the strengths of various low-level models."], "abstract": "Car-following models have made significant contributions to our understanding of longitudinal driving behavior. However, they often exhibit limited accuracy and flexibility, as they cannot fully capture the complexity inherent in car-following processes, or may falter in unseen scenarios due to their reliance on confined driving skills present in training data. It is worth noting that each car-following model possesses its own strengths and weaknesses depending on specific driving scenarios. Therefore, we propose EnsembleFollower, a hierarchical planning framework for achieving advanced human-like car-following. The EnsembleFollower framework involves a high-level Reinforcement Learning-based agent responsible for judiciously managing multiple low-level car-following models according to the current state, either by selecting an appropriate low-level model to perform an action or by allocating different weights across all low-level components. Moreover, we propose a jerk-constrained kinematic model for more convincing car-following simulations. We evaluate the proposed method based on real-world driving data from the HighD dataset. The experimental results illustrate that EnsembleFollower yields improved accuracy of human-like behavior and achieves effectiveness in combining hybrid models, demonstrating that our proposed framework can handle diverse car-following conditions by leveraging the strengths of various low-level models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ec6bb7c3fd7ad9c7", "title": "Iterative Reward Shaping using Human Feedback for Correcting Reward Misspecification", "authors": ["Jasmina Gajcin", "James McCarthy", "Rahul Nair", "Radu Marinescu", "Elizabeth Daly", "Ivana Dusparic"], "year": 2023, "published": "2023-08-30T11:45:40Z", "updated": "2023-08-30T11:45:40Z", "arxiv_id": "http://arxiv.org/abs/2308.15969v1", "url": "https://arxiv.org/abs/2308.15969v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["A well-defined reward function is crucial for successful training of an reinforcement learning (RL) agent."], "abstract": "A well-defined reward function is crucial for successful training of an reinforcement learning (RL) agent. However, defining a suitable reward function is a notoriously challenging task, especially in complex, multi-objective environments. Developers often have to resort to starting with an initial, potentially misspecified reward function, and iteratively adjusting its parameters, based on observed learned behavior. In this work, we aim to automate this process by proposing ITERS, an iterative reward shaping approach using human feedback for mitigating the effects of a misspecified reward function. Our approach allows the user to provide trajectory-level feedback on agent's behavior during training, which can be integrated as a reward shaping signal in the following training iteration. We also allow the user to provide explanations of their feedback, which are used to augment the feedback and reduce user effort and feedback frequency. We evaluate ITERS in three environments and show that it can successfully correct misspecified reward functions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5e3b6aac7d78012d", "title": "Cyclophobic Reinforcement Learning", "authors": ["Stefan Sylvius Wagner", "Peter Arndt", "Jan Robine", "Stefan Harmeling"], "year": 2023, "published": "2023-08-30T09:38:44Z", "updated": "2023-08-30T09:38:44Z", "arxiv_id": "http://arxiv.org/abs/2308.15911v1", "url": "https://arxiv.org/abs/2308.15911v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.RO"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["Augmenting the cyclophobic intrinsic reward with a sequence of hierarchical representations based on the agent's cropped observations we are able to achieve excellent results in the MiniGrid and MiniHack environments."], "abstract": "In environments with sparse rewards, finding a good inductive bias for exploration is crucial to the agent's success. However, there are two competing goals: novelty search and systematic exploration. While existing approaches such as curiosity-driven exploration find novelty, they sometimes do not systematically explore the whole state space, akin to depth-first-search vs breadth-first-search. In this paper, we propose a new intrinsic reward that is cyclophobic, i.e., it does not reward novelty, but punishes redundancy by avoiding cycles. Augmenting the cyclophobic intrinsic reward with a sequence of hierarchical representations based on the agent's cropped observations we are able to achieve excellent results in the MiniGrid and MiniHack environments. Both are particularly hard, as they require complex interactions with different objects in order to be solved. Detailed comparisons with previous approaches and thorough ablation studies show that our newly proposed cyclophobic reinforcement learning is more sample efficient than other state of the art methods in a variety of tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bf803b4888ef8e7f", "title": "Beyond Traditional Neural Networks: Toward adding Reasoning and Learning Capabilities through Computational Logic Techniques", "authors": ["Andrea Rafanelli"], "year": 2023, "published": "2023-08-30T09:09:42Z", "updated": "2023-08-30T09:09:42Z", "arxiv_id": "http://arxiv.org/abs/2308.15899v1", "url": "https://arxiv.org/abs/2308.15899v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG", "cs.LO", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["This work proposes solutions to improve the knowledge injection process and integrate elements of ML and logic into multi-agent systems (MAS)."], "abstract": "Deep Learning (DL) models have become popular for solving complex problems, but they have limitations such as the need for high-quality training data, lack of transparency, and robustness issues. Neuro-Symbolic AI has emerged as a promising approach combining the strengths of neural networks and symbolic reasoning. Symbolic knowledge injection (SKI) techniques are a popular method to incorporate symbolic knowledge into sub-symbolic systems. This work proposes solutions to improve the knowledge injection process and integrate elements of ML and logic into multi-agent systems (MAS).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-eb41effd3fed6b79", "title": "Penalization Framework For Autonomous Agents Using Answer Set Programming", "authors": ["Vineel S. K. Tummala"], "year": 2023, "published": "2023-08-30T09:09:27Z", "updated": "2023-08-30T09:09:27Z", "arxiv_id": "http://arxiv.org/abs/2309.04487v1", "url": "https://arxiv.org/abs/2309.04487v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LO"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["This paper presents a framework for enforcing penalties on intelligent agents that do not comply with authorization or obligation policies in a changing environment."], "abstract": "This paper presents a framework for enforcing penalties on intelligent agents that do not comply with authorization or obligation policies in a changing environment. A framework is proposed to represent and reason about penalties in plans, and an algorithm is proposed to penalize an agent's actions based on their level of compliance with respect to authorization and obligation policies. Being aware of penalties an agent can choose a plan with a minimal total penalty, unless there is an emergency goal like saving a human's life. The paper concludes that this framework can reprimand insubordinate agents.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-a93435fcc051d346", "title": "(Mis)align: A Simple Dynamic Framework for Modeling Interpersonal Coordination", "authors": ["Grace Qiyuan Miao", "Rick Dale", "Alexia Galati"], "year": 2023, "published": "2023-08-30T08:55:30Z", "updated": "2023-08-30T08:55:30Z", "arxiv_id": "http://arxiv.org/abs/2308.15864v1", "url": "https://arxiv.org/abs/2308.15864v1", "primary_category": "cs.MA", "categories": ["cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["As people coordinate in daily interactions, they engage in different patterns of behavior to achieve successful outcomes."], "abstract": "As people coordinate in daily interactions, they engage in different patterns of behavior to achieve successful outcomes. This includes both synchrony - the temporal coordination of the same behaviors at the same time - and complementarity - the coordination of the same or different behaviors that may occur at different relative times. Using computational methods, we develop a simple framework to describe the interpersonal dynamics of behavioral synchrony and complementarity over time, and explore their task dependence. A key feature of this framework is the inclusion of a task context that mediates interactions, and consists of active, inactive, and inhibitory constraints on communication. Initial simulation results show that these task constraints can be a robust predictor of simulated agents' behaviors over time. We also show that the framework can reproduce some general patterns observed in human interaction data. We describe preliminary theoretical implications from these results, and relate them to broader proposals of synergistic self-organization in communication.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-306b612fc2001a70", "title": "Benchmarking Robustness and Generalization in Multi-Agent Systems: A Case Study on Neural MMO", "authors": ["Yangkun Chen", "Joseph Suarez", "Junjie Zhang", "Chenghui Yu", "Bo Wu", "Hanmo Chen", "Hengman Zhu", "Rui Du", "Shanliang Qian", "Shuai Liu", "Weijun Hong", "Jinke He", "Yibing Zhang", "Liang Zhao", "Clare Zhu", "Julian Togelius", "Sharada Mohanty", "Jiaxin Chen", "Xiu Li", "Xiaolong Zhu", "Phillip Isola"], "year": 2023, "published": "2023-08-30T07:16:11Z", "updated": "2023-08-30T07:16:11Z", "arxiv_id": "http://arxiv.org/abs/2308.15802v1", "url": "https://arxiv.org/abs/2308.15802v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["This competition targets robustness and generalization in multi-agent systems: participants train teams of agents to complete a multi-task objective against opponents not seen during training."], "abstract": "We present the results of the second Neural MMO challenge, hosted at IJCAI 2022, which received 1600+ submissions. This competition targets robustness and generalization in multi-agent systems: participants train teams of agents to complete a multi-task objective against opponents not seen during training. The competition combines relatively complex environment design with large numbers of agents in the environment. The top submissions demonstrate strong success on this task using mostly standard reinforcement learning (RL) methods combined with domain-specific engineering. We summarize the competition design and results and suggest that, as an academic community, competitions may be a powerful approach to solving hard problems and establishing a solid benchmark for algorithms. We will open-source our benchmark including the environment wrapper, baselines, a visualization tool, and selected policies for further research.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5b8a0fd1bd5a8ca6", "title": "Deep Reinforcement Learning Based Framework for Mobile Energy Disseminator Dispatching to Charge On-the-Road Electric Vehicles", "authors": ["Jiaming Wang", "Jiqian Dong", "Sikai Chen", "Shreyas Sundaram", "Samuel Labi"], "year": 2023, "published": "2023-08-29T22:23:52Z", "updated": "2023-08-29T22:23:52Z", "arxiv_id": "http://arxiv.org/abs/2308.15656v1", "url": "https://arxiv.org/abs/2308.15656v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["The exponential growth of electric vehicles (EVs) presents novel challenges in preserving battery health and in addressing the persistent problem of vehicle range anxiety."], "abstract": "The exponential growth of electric vehicles (EVs) presents novel challenges in preserving battery health and in addressing the persistent problem of vehicle range anxiety. To address these concerns, wireless charging, particularly, Mobile Energy Disseminators (MEDs) have emerged as a promising solution. The MED is mounted behind a large vehicle and charges all participating EVs within a radius upstream of it. Unfortuantely, during such V2V charging, the MED and EVs inadvertently form platoons, thereby occupying multiple lanes and impairing overall corridor travel efficiency. In addition, constrained budgets for MED deployment necessitate the development of an effective dispatching strategy to determine optimal timing and locations for introducing the MEDs into traffic. This paper proposes a deep reinforcement learning (DRL) based methodology to develop a vehicle dispatching framework. In the first component of the framework, we develop a realistic reinforcement learning environment termed \"ChargingEnv\" which incorporates a reliable charging simulation system that accounts for common practical issues in wireless charging deployment, specifically, the charging panel misalignment. The second component, the Proximal-Policy Optimization (PPO) agent, is trained to control MED dispatching through continuous interactions with ChargingEnv. Numerical experiments were carried out to demonstrate the demonstrate the efficacy of the proposed MED deployment decision processor. The experiment results suggest that the proposed model can significantly enhance EV travel range while efficiently deploying a optimal number of MEDs. The proposed model is found to be not only practical in its applicability but also has promises of real-world effectiveness. The proposed model can help travelers to maximize EV range and help road agencies or private-sector vendors to manage the deployment of MEDs efficiently.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6eba50d104efc1de", "title": "Adversarial Style Transfer for Robust Policy Optimization in Deep Reinforcement Learning", "authors": ["Md Masudur Rahman", "Yexiang Xue"], "year": 2023, "published": "2023-08-29T18:17:35Z", "updated": "2023-08-29T18:17:35Z", "arxiv_id": "http://arxiv.org/abs/2308.15550v1", "url": "https://arxiv.org/abs/2308.15550v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["ying"], "key_results": ["This paper proposes an algorithm that aims to improve generalization for reinforcement learning agents by removing overfitting to confounding features.", "Empirically, ARPO shows improved performance compared to a few baseline algorithms, including data augmentation."], "abstract": "This paper proposes an algorithm that aims to improve generalization for reinforcement learning agents by removing overfitting to confounding features. Our approach consists of a max-min game theoretic objective. A generator transfers the style of observation during reinforcement learning. An additional goal of the generator is to perturb the observation, which maximizes the agent's probability of taking a different action. In contrast, a policy network updates its parameters to minimize the effect of such perturbations, thus staying robust while maximizing the expected future reward. Based on this setup, we propose a practical deep reinforcement learning algorithm, Adversarial Robust Policy Optimization (ARPO), to find a robust policy that generalizes to unseen environments. We evaluate our approach on Procgen and Distracting Control Suite for generalization and sample efficiency. Empirically, ARPO shows improved performance compared to a few baseline algorithms, including data augmentation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5e1e50896cf44762", "title": "Decentralized Multi-agent Reinforcement Learning based State-of-Charge Balancing Strategy for Distributed Energy Storage System", "authors": ["Zheng Xiong", "Biao Luo", "Bing-Chuan Wang", "Xiaodong Xu", "Xiaodong Liu", "Tingwen Huang"], "year": 2023, "published": "2023-08-29T15:48:49Z", "updated": "2023-08-29T15:48:49Z", "arxiv_id": "http://arxiv.org/abs/2308.15394v1", "url": "https://arxiv.org/abs/2308.15394v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG", "eess.SY"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["This paper develops a Decentralized Multi-Agent Reinforcement Learning (Dec-MARL) method to solve the SoC balancing problem in the distributed energy storage system (DESS)."], "abstract": "This paper develops a Decentralized Multi-Agent Reinforcement Learning (Dec-MARL) method to solve the SoC balancing problem in the distributed energy storage system (DESS). First, the SoC balancing problem is formulated into a finite Markov decision process with action constraints derived from demand balance, which can be solved by Dec-MARL. Specifically, the first-order average consensus algorithm is utilized to expand the observations of the DESS state in a fully-decentralized way, and the initial actions (i.e., output power) are decided by the agents (i.e., energy storage units) according to these observations. In order to get the final actions in the allowable range, a counterfactual demand balance algorithm is proposed to balance the total demand and the initial actions. Next, the agents execute the final actions and get local rewards from the environment, and the DESS steps into the next state. Finally, through the first-order average consensus algorithm, the agents get the average reward and the expended observation of the next state for later training. By the above procedure, Dec-MARL reveals outstanding performance in a fully-decentralized system without any expert experience or constructing any complicated model. Besides, it is flexible and can be extended to other decentralized multi-agent systems straightforwardly. Extensive simulations have validated the effectiveness and efficiency of Dec-MARL.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-3028c5be92c892d2", "title": "Bearing-based Formation with Disturbance Rejection", "authors": ["Haoshu Cheng", "Jie Huang"], "year": 2023, "published": "2023-08-29T12:40:04Z", "updated": "2023-08-29T12:40:04Z", "arxiv_id": "http://arxiv.org/abs/2308.15260v1", "url": "https://arxiv.org/abs/2308.15260v1", "primary_category": "eess.SY", "categories": ["eess.SY", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["This paper considers the problem of the bearing-based formation control with disturbance rejection for a group of agents under the leader-follower structure."], "abstract": "This paper considers the problem of the bearing-based formation control with disturbance rejection for a group of agents under the leader-follower structure. The disturbances are in the form of a trigonometric polynomial with arbitrary unknown amplitudes, unknown initial phases, and known or unknown frequencies. For the case of the known frequencies, we employ the canonical internal model to solve the problem, and, for the case of the unknown frequencies, we combine the canonical internal model and {some} distributed adaptive control technique to deal with the problem. It is noted that the existing results can only handle constant input disturbances by continuous control laws or disturbances with known {bounds} by discontinuous control laws. The first case is a special case of our result. The second case cannot cover our results because the bound of our disturbance is unknown. Moreover, our control law is smooth.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0974d74369c6511d", "title": "From DDMs to DNNs: Using process data and models of decision-making to improve human-AI interactions", "authors": ["Mrugsen Nagsen Gopnarayan", "Jaan Aru", "Sebastian Gluth"], "year": 2023, "published": "2023-08-29T11:27:22Z", "updated": "2025-07-23T11:02:12Z", "arxiv_id": "http://arxiv.org/abs/2308.15225v3", "url": "https://arxiv.org/abs/2308.15225v3", "primary_category": "q-bio.NC", "categories": ["q-bio.NC", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Here, we argue that artificial intelligence (AI) research would benefit from a stronger focus on insights about how decisions emerge over time and incorporate related process data to improve AI predictions in general and human-AI interactions in particular.", "Finally, we outline how a more principled inclusion of the evidence-accumulation framework into the training and use of AI can help to improve human-AI interactions in the future."], "abstract": "Over the past decades, cognitive neuroscientists and behavioral economists have recognized the value of describing the process of decision making in detail and modeling the emergence of decisions over time. For example, the time it takes to decide can reveal more about an agent's true hidden preferences than only the decision itself. Similarly, data that track the ongoing decision process such as eye movements or neural recordings contain critical information that can be exploited, even if no decision is made. Here, we argue that artificial intelligence (AI) research would benefit from a stronger focus on insights about how decisions emerge over time and incorporate related process data to improve AI predictions in general and human-AI interactions in particular. First, we introduce a highly established computational framework that assumes decisions to emerge from the noisy accumulation of evidence, and we present related empirical work in psychology, neuroscience, and economics. Next, we discuss to what extent current approaches in multi-agent AI do or do not incorporate process data and models of decision making. Finally, we outline how a more principled inclusion of the evidence-accumulation framework into the training and use of AI can help to improve human-AI interactions in the future.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-396614c4ebe7db5a", "title": "FurChat: An Embodied Conversational Agent using LLMs, Combining Open and Closed-Domain Dialogue with Facial Expressions", "authors": ["Neeraj Cherakara", "Finny Varghese", "Sheena Shabana", "Nivan Nelson", "Abhiram Karukayil", "Rohith Kulothungan", "Mohammed Afil Farhan", "Birthe Nesset", "Meriam Moujahid", "Tanvi Dinkar", "Verena Rieser", "Oliver Lemon"], "year": 2023, "published": "2023-08-29T11:08:40Z", "updated": "2023-08-30T13:13:19Z", "arxiv_id": "http://arxiv.org/abs/2308.15214v2", "url": "https://arxiv.org/abs/2308.15214v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.HC", "cs.RO"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["GPT-3.5"], "key_results": ["The system utilises the state-of-the-art GPT-3.5 model to generate such information along with domain-general conversations and facial expressions based on prompt engineering."], "abstract": "We demonstrate an embodied conversational agent that can function as a receptionist and generate a mixture of open and closed-domain dialogue along with facial expressions, by using a large language model (LLM) to develop an engaging conversation. We deployed the system onto a Furhat robot, which is highly expressive and capable of using both verbal and nonverbal cues during interaction. The system was designed specifically for the National Robotarium to interact with visitors through natural conversations, providing them with information about the facilities, research, news, upcoming events, etc. The system utilises the state-of-the-art GPT-3.5 model to generate such information along with domain-general conversations and facial expressions based on prompt engineering.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-32826feae94f7a46", "title": "LTLf Best-Effort Synthesis in Nondeterministic Planning Domains", "authors": ["Giuseppe De Giacomo", "Gianmarco Parretti", "Shufang Zhu"], "year": 2023, "published": "2023-08-29T10:10:41Z", "updated": "2023-08-29T10:10:41Z", "arxiv_id": "http://arxiv.org/abs/2308.15188v1", "url": "https://arxiv.org/abs/2308.15188v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.FL", "cs.GT", "cs.RO"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["The notion of best-effort strategy has been introduced to also deal with the scenario when no agent strategy exists that fulfills the goal against every possible nondeterministic environment reaction."], "abstract": "We study best-effort strategies (aka plans) in fully observable nondeterministic domains (FOND) for goals expressed in Linear Temporal Logic on Finite Traces (LTLf). The notion of best-effort strategy has been introduced to also deal with the scenario when no agent strategy exists that fulfills the goal against every possible nondeterministic environment reaction. Such strategies fulfill the goal if possible, and do their best to do so otherwise. We present a game-theoretic technique for synthesizing best-effort strategies that exploit the specificity of nondeterministic planning domains. We formally show its correctness and demonstrate its effectiveness experimentally, exhibiting a much greater scalability with respect to a direct best-effort synthesis approach based on re-expressing the planning domain as generic environment specifications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6f1a7752b9fa3e7a", "title": "eDKM: An Efficient and Accurate Train-time Weight Clustering for Large Language Models", "authors": ["Minsik Cho", "Keivan A. Vahid", "Qichen Fu", "Saurabh Adya", "Carlo C Del Mundo", "Mohammad Rastegari", "Devang Naik", "Peter Zatloukal"], "year": 2023, "published": "2023-09-02T15:16:35Z", "updated": "2023-09-13T19:17:37Z", "arxiv_id": "http://arxiv.org/abs/2309.00964v2", "url": "https://arxiv.org/abs/2309.00964v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Memory / RAG"], "benchmarks": [], "models": ["LLaMA 7B", "ying"], "key_results": ["Especially, Differentiable KMeans Clustering, or DKM, has shown the state-of-the-art trade-off between compression ratio and accuracy regression, but its large memory complexity makes it nearly impossible to apply to train-time LLM compression."], "abstract": "Since Large Language Models or LLMs have demonstrated high-quality performance on many complex language tasks, there is a great interest in bringing these LLMs to mobile devices for faster responses and better privacy protection. However, the size of LLMs (i.e., billions of parameters) requires highly effective compression to fit into storage-limited devices. Among many compression techniques, weight-clustering, a form of non-linear quantization, is one of the leading candidates for LLM compression, and supported by modern smartphones. Yet, its training overhead is prohibitively significant for LLM fine-tuning. Especially, Differentiable KMeans Clustering, or DKM, has shown the state-of-the-art trade-off between compression ratio and accuracy regression, but its large memory complexity makes it nearly impossible to apply to train-time LLM compression. In this paper, we propose a memory-efficient DKM implementation, eDKM powered by novel techniques to reduce the memory footprint of DKM by orders of magnitudes. For a given tensor to be saved on CPU for the backward pass of DKM, we compressed the tensor by applying uniquification and sharding after checking if there is no duplicated tensor previously copied to CPU. Our experimental results demonstrate that \\prjname can fine-tune and compress a pretrained LLaMA 7B model from 12.6 GB to 2.5 GB (3bit/weight) with the Alpaca dataset by reducing the train-time memory footprint of a decoder layer by 130$\\times$, while delivering good accuracy on broader LLM benchmarks (i.e., 77.7% for PIQA, 66.1% for Winograde, and so on).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-795ec2d8506310b8", "title": "Structured Radial Basis Function Network: Modelling Diversity for Multiple Hypotheses Prediction", "authors": ["Alejandro Rodriguez Dominguez", "Muhammad Shahzad", "Xia Hong"], "year": 2023, "published": "2023-09-02T01:27:53Z", "updated": "2024-09-20T10:11:26Z", "arxiv_id": "http://arxiv.org/abs/2309.00781v2", "url": "https://arxiv.org/abs/2309.00781v2", "primary_category": "cs.LG", "categories": ["cs.LG", "stat.ML"], "pillar": "Efficient Training & Inference", "methods": [], "benchmarks": [], "models": [], "key_results": ["Multi-modal problems can be effectively addressed using multiple hypothesis frameworks, but integrating these frameworks into learning models poses significant challenges."], "abstract": "Multi-modal problems can be effectively addressed using multiple hypothesis frameworks, but integrating these frameworks into learning models poses significant challenges. This paper introduces a Structured Radial Basis Function Network (s-RBFN) as an ensemble of multiple hypothesis predictors for regression. During the training of the predictors, first the centroidal Voronoi tessellations are formed based on their losses and the true labels, representing geometrically the set of multiple hypotheses. Then, the trained predictors are used to compute a structured dataset with their predictions, including centers and scales for the basis functions. A radial basis function network, with each basis function focused on a particular hypothesis, is subsequently trained using this structured dataset for multiple hypotheses prediction. The s-RBFN is designed to train efficiently while controlling diversity in ensemble learning parametrically. The least-squares approach for training the structured ensemble model provides a closed-form solution for multiple hypotheses and structured predictions. During the formation of the structured dataset, a parameter is employed to avoid mode collapse by controlling tessellation shapes. This parameter provides a mechanism to balance diversity and generalization performance for the s-RBFN. The empirical validation on two multivariate prediction datasets-air quality and energy appliance predictions-demonstrates the superior generalization performance and computational efficiency of the structured ensemble model compared to other models and their single-hypothesis counterparts.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-61e26a7a7c8123ec", "title": "PathLDM: Text conditioned Latent Diffusion Model for Histopathology", "authors": ["Srikar Yellapragada", "Alexandros Graikos", "Prateek Prasanna", "Tahsin Kurc", "Joel Saltz", "Dimitris Samaras"], "year": 2023, "published": "2023-09-01T22:08:32Z", "updated": "2023-11-30T20:20:23Z", "arxiv_id": "http://arxiv.org/abs/2309.00748v2", "url": "https://arxiv.org/abs/2309.00748v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["To achieve high-quality results, diffusion models must be trained on large datasets.", "Through strategic conditioning and necessary architectural enhancements, we achieved a SoTA FID score of 7.64 for text-to-image generation on the TCGA-BRCA dataset, significantly outperforming the closest text-conditioned competitor with FID 30.1."], "abstract": "To achieve high-quality results, diffusion models must be trained on large datasets. This can be notably prohibitive for models in specialized domains, such as computational pathology. Conditioning on labeled data is known to help in data-efficient model training. Therefore, histopathology reports, which are rich in valuable clinical information, are an ideal choice as guidance for a histopathology generative model. In this paper, we introduce PathLDM, the first text-conditioned Latent Diffusion Model tailored for generating high-quality histopathology images. Leveraging the rich contextual information provided by pathology text reports, our approach fuses image and textual data to enhance the generation process. By utilizing GPT's capabilities to distill and summarize complex text reports, we establish an effective conditioning mechanism. Through strategic conditioning and necessary architectural enhancements, we achieved a SoTA FID score of 7.64 for text-to-image generation on the TCGA-BRCA dataset, significantly outperforming the closest text-conditioned competitor with FID 30.1.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6b833b6b6eb558bb", "title": "BatchPrompt: Accomplish more with less", "authors": ["Jianzhe Lin", "Maurice Diesendruck", "Liang Du", "Robin Abraham"], "year": 2023, "published": "2023-09-01T10:44:36Z", "updated": "2024-07-15T05:42:34Z", "arxiv_id": "http://arxiv.org/abs/2309.00384v3", "url": "https://arxiv.org/abs/2309.00384v3", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Long Context"], "benchmarks": [], "models": ["GPT-4", "gpt-3.5-turbo"], "key_results": ["Our comprehensive experimental evaluation demonstrates that BPE can boost the performance of BatchPrompt with a striking margin on a range of popular NLP tasks, including question answering (Boolq), textual entailment (RTE), and duplicate questions identification (QQP).", "To the best of our knowledge, this is the first work to technically improve prompting efficiency of large language models."], "abstract": "As the ever-increasing token limits of large language models (LLMs) have enabled long context as input, prompting with single data samples might no longer an efficient way. A straightforward strategy improving efficiency is to batch data within the token limit (e.g., 8k for gpt-3.5-turbo; 32k for GPT-4), which we call BatchPrompt. We have two initial observations for prompting with batched data. First, we find that prompting with batched data in longer contexts will inevitably lead to worse performance, compared to single-data prompting. Second, the performance of the language model is significantly correlated with the positions and order of the batched data, due to the corresponding change in decoder context. To retain efficiency and overcome performance loss, we propose Batch Permutation and Ensembling (BPE), and a novel Self-reflection-guided EArly Stopping (SEAS) technique. Our comprehensive experimental evaluation demonstrates that BPE can boost the performance of BatchPrompt with a striking margin on a range of popular NLP tasks, including question answering (Boolq), textual entailment (RTE), and duplicate questions identification (QQP). These performances are even competitive with/higher than single-data prompting(SinglePrompt), while BatchPrompt requires much fewer LLM calls and input tokens (For SinglePrompt v.s. BatchPrompt with batch size 32, using just 9%-16% the number of LLM calls, Boolq accuracy 90.6% to 90.9% with 27.4% tokens, QQP accuracy 87.2% to 88.4% with 18.6% tokens, RTE accuracy 91.5% to 91.1% with 30.8% tokens). To the best of our knowledge, this is the first work to technically improve prompting efficiency of large language models. We hope our simple yet effective approach will shed light on the future research of large language models. The code will be released.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-55e098dc2c7cf424", "title": "Learning multi-modal generative models with permutation-invariant encoders and tighter variational objectives", "authors": ["Marcel Hirt", "Domenico Campolo", "Victoria Leong", "Juan-Pablo Ortega"], "year": 2023, "published": "2023-09-01T10:32:21Z", "updated": "2024-09-24T13:59:59Z", "arxiv_id": "http://arxiv.org/abs/2309.00380v3", "url": "https://arxiv.org/abs/2309.00380v3", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["MoE"], "benchmarks": [], "models": ["yield"], "key_results": ["Devising deep latent variable models for multi-modal data has been a long-standing theme in machine learning research."], "abstract": "Devising deep latent variable models for multi-modal data has been a long-standing theme in machine learning research. Multi-modal Variational Autoencoders (VAEs) have been a popular generative model class that learns latent representations that jointly explain multiple modalities. Various objective functions for such models have been suggested, often motivated as lower bounds on the multi-modal data log-likelihood or from information-theoretic considerations. To encode latent variables from different modality subsets, Product-of-Experts (PoE) or Mixture-of-Experts (MoE) aggregation schemes have been routinely used and shown to yield different trade-offs, for instance, regarding their generative quality or consistency across multiple modalities. In this work, we consider a variational objective that can tightly approximate the data log-likelihood. We develop more flexible aggregation schemes that avoid the inductive biases in PoE or MoE approaches by combining encoded features from different modalities based on permutation-invariant neural networks. Our numerical experiments illustrate trade-offs for multi-modal variational objectives and various aggregation schemes. We show that our variational objective and more flexible aggregation models can become beneficial when one wants to approximate the true joint distribution over observed modalities and latent variables in identifiable models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-189fc7af8c1c2e91", "title": "RepCodec: A Speech Representation Codec for Speech Tokenization", "authors": ["Zhichao Huang", "Chutong Meng", "Tom Ko"], "year": 2023, "published": "2023-08-31T23:26:10Z", "updated": "2024-07-22T09:53:44Z", "arxiv_id": "http://arxiv.org/abs/2309.00169v3", "url": "https://arxiv.org/abs/2309.00169v3", "primary_category": "eess.AS", "categories": ["eess.AS", "cs.LG", "cs.SD"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["BERT"], "key_results": ["To improve the performance of these discrete speech tokens, we present RepCodec, a novel speech representation codec for semantic speech tokenization.", "The extensive experiments illustrate that RepCodec, by virtue of its enhanced information retention capacity, significantly outperforms the widely used k-means clustering approach in both speech understanding and generation."], "abstract": "With recent rapid growth of large language models (LLMs), discrete speech tokenization has played an important role for injecting speech into LLMs. However, this discretization gives rise to a loss of information, consequently impairing overall performance. To improve the performance of these discrete speech tokens, we present RepCodec, a novel speech representation codec for semantic speech tokenization. In contrast to audio codecs which reconstruct the raw audio, RepCodec learns a vector quantization codebook through reconstructing speech representations from speech encoders like HuBERT or data2vec. Together, the speech encoder, the codec encoder and the vector quantization codebook form a pipeline for converting speech waveforms into semantic tokens. The extensive experiments illustrate that RepCodec, by virtue of its enhanced information retention capacity, significantly outperforms the widely used k-means clustering approach in both speech understanding and generation. Furthermore, this superiority extends across various speech encoders and languages, affirming the robustness of RepCodec. We believe our method can facilitate large language modeling research on speech processing.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-35bb386d9aafd0e0", "title": "QS-TTS: Towards Semi-Supervised Text-to-Speech Synthesis via Vector-Quantized Self-Supervised Speech Representation Learning", "authors": ["Haohan Guo", "Fenglong Xie", "Jiawen Kang", "Yujia Xiao", "Xixin Wu", "Helen Meng"], "year": 2023, "published": "2023-08-31T20:25:44Z", "updated": "2023-08-31T20:25:44Z", "arxiv_id": "http://arxiv.org/abs/2309.00126v1", "url": "https://arxiv.org/abs/2309.00126v1", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.CL", "eess.AS"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["This paper proposes a novel semi-supervised TTS framework, QS-TTS, to improve TTS quality with lower supervised data requirements via Vector-Quantized Self-Supervised Speech Representation Learning (VQ-S3RL) utilizing more unlabeled speech audio.", "Moreover, comparing various speech representations and transfer learning methods in TTS further validates the notable improvement of the proposed VQ-S3RL to TTS, showing the best audio quality and intelligibility metrics."], "abstract": "This paper proposes a novel semi-supervised TTS framework, QS-TTS, to improve TTS quality with lower supervised data requirements via Vector-Quantized Self-Supervised Speech Representation Learning (VQ-S3RL) utilizing more unlabeled speech audio. This framework comprises two VQ-S3R learners: first, the principal learner aims to provide a generative Multi-Stage Multi-Codebook (MSMC) VQ-S3R via the MSMC-VQ-GAN combined with the contrastive S3RL, while decoding it back to the high-quality audio; then, the associate learner further abstracts the MSMC representation into a highly-compact VQ representation through a VQ-VAE. These two generative VQ-S3R learners provide profitable speech representations and pre-trained models for TTS, significantly improving synthesis quality with the lower requirement for supervised data. QS-TTS is evaluated comprehensively under various scenarios via subjective and objective tests in experiments. The results powerfully demonstrate the superior performance of QS-TTS, winning the highest MOS over supervised or semi-supervised baseline TTS approaches, especially in low-resource scenarios. Moreover, comparing various speech representations and transfer learning methods in TTS further validates the notable improvement of the proposed VQ-S3RL to TTS, showing the best audio quality and intelligibility metrics. The trend of slower decay in the synthesis quality of QS-TTS with decreasing supervised data further highlights its lower requirements for supervised data, indicating its great potential in low-resource scenarios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-15cfb4a8412528f1", "title": "GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields", "authors": ["Yanjie Ze", "Ge Yan", "Yueh-Hua Wu", "Annabella Macaluso", "Yuying Ge", "Jianglong Ye", "Nicklas Hansen", "Li Erran Li", "Xiaolong Wang"], "year": 2023, "published": "2023-08-31T17:52:10Z", "updated": "2024-07-28T02:10:19Z", "arxiv_id": "http://arxiv.org/abs/2308.16891v3", "url": "https://arxiv.org/abs/2308.16891v3", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Video Generation", "VLM"], "benchmarks": [], "models": ["Stable Diffusion"], "key_results": ["To achieve this goal, the robot needs to have a comprehensive understanding of the 3D structure and semantics of the scene.", "We observe a substantial improvement of GNFactor over current state-of-the-art methods in seen and unseen tasks, demonstrating the strong generalization ability of GNFactor."], "abstract": "It is a long-standing problem in robotics to develop agents capable of executing diverse manipulation tasks from visual observations in unstructured real-world environments. To achieve this goal, the robot needs to have a comprehensive understanding of the 3D structure and semantics of the scene. In this work, we present $\\textbf{GNFactor}$, a visual behavior cloning agent for multi-task robotic manipulation with $\\textbf{G}$eneralizable $\\textbf{N}$eural feature $\\textbf{F}$ields. GNFactor jointly optimizes a generalizable neural field (GNF) as a reconstruction module and a Perceiver Transformer as a decision-making module, leveraging a shared deep 3D voxel representation. To incorporate semantics in 3D, the reconstruction module utilizes a vision-language foundation model ($\\textit{e.g.}$, Stable Diffusion) to distill rich semantic information into the deep 3D voxel. We evaluate GNFactor on 3 real robot tasks and perform detailed ablations on 10 RLBench tasks with a limited number of demonstrations. We observe a substantial improvement of GNFactor over current state-of-the-art methods in seen and unseen tasks, demonstrating the strong generalization ability of GNFactor. Our project website is https://yanjieze.com/GNFactor/ .", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c735bbe83f6e8528", "title": "SpeechTokenizer: Unified Speech Tokenizer for Speech Large Language Models", "authors": ["Xin Zhang", "Dong Zhang", "Shimin Li", "Yaqian Zhou", "Xipeng Qiu"], "year": 2023, "published": "2023-08-31T12:53:09Z", "updated": "2024-01-23T01:56:57Z", "arxiv_id": "http://arxiv.org/abs/2308.16692v2", "url": "https://arxiv.org/abs/2308.16692v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.SD", "eess.AS"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["ying"], "key_results": ["Also, USLM outperforms VALL-E in zero-shot Text-to-Speech tasks."], "abstract": "Current speech large language models build upon discrete speech representations, which can be categorized into semantic tokens and acoustic tokens. However, existing speech tokens are not specifically designed for speech language modeling. To assess the suitability of speech tokens for building speech language models, we established the first benchmark, SLMTokBench. Our results indicate that neither semantic nor acoustic tokens are ideal for this purpose. Therefore, we propose SpeechTokenizer, a unified speech tokenizer for speech large language models. SpeechTokenizer adopts the Encoder-Decoder architecture with residual vector quantization (RVQ). Unifying semantic and acoustic tokens, SpeechTokenizer disentangles different aspects of speech information hierarchically across different RVQ layers. Furthermore, We construct a Unified Speech Language Model (USLM) leveraging SpeechTokenizer. Experiments show that SpeechTokenizer performs comparably to EnCodec in speech reconstruction and demonstrates strong performance on the SLMTokBench benchmark. Also, USLM outperforms VALL-E in zero-shot Text-to-Speech tasks. Code and models are available at https://github.com/ZhangXInFD/SpeechTokenizer/.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1b43ed239a9fd12c", "title": "Continual Learning From a Stream of APIs", "authors": ["Enneng Yang", "Zhenyi Wang", "Li Shen", "Nan Yin", "Tongliang Liu", "Guibing Guo", "Xingwei Wang", "Dacheng Tao"], "year": 2023, "published": "2023-08-31T11:16:00Z", "updated": "2024-09-12T08:34:10Z", "arxiv_id": "http://arxiv.org/abs/2309.00023v2", "url": "https://arxiv.org/abs/2309.00023v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": ["ImageNet"], "models": ["R10", "R100", "phic", "ying"], "key_results": ["This paper considers two practical-yet-novel CL settings: data-efficient CL (DECL-APIs) and data-free CL (DFCL-APIs), which achieve CL from a stream of APIs with partial or no raw data.", "In the DECL-APIs setting, our method achieves 0.97x, 0.75x and 0.69x performance of classic CL on CIFAR10, CIFAR100, and MiniImageNet."], "abstract": "Continual learning (CL) aims to learn new tasks without forgetting previous tasks. However, existing CL methods require a large amount of raw data, which is often unavailable due to copyright considerations and privacy risks. Instead, stakeholders usually release pre-trained machine learning models as a service (MLaaS), which users can access via APIs. This paper considers two practical-yet-novel CL settings: data-efficient CL (DECL-APIs) and data-free CL (DFCL-APIs), which achieve CL from a stream of APIs with partial or no raw data. Performing CL under these two new settings faces several challenges: unavailable full raw data, unknown model parameters, heterogeneous models of arbitrary architecture and scale, and catastrophic forgetting of previous APIs. To overcome these issues, we propose a novel data-free cooperative continual distillation learning framework that distills knowledge from a stream of APIs into a CL model by generating pseudo data, just by querying APIs. Specifically, our framework includes two cooperative generators and one CL model, forming their training as an adversarial game. We first use the CL model and the current API as fixed discriminators to train generators via a derivative-free method. Generators adversarially generate hard and diverse synthetic data to maximize the response gap between the CL model and the API. Next, we train the CL model by minimizing the gap between the responses of the CL model and the black-box API on synthetic data, to transfer the API's knowledge to the CL model. Furthermore, we propose a new regularization term based on network similarity to prevent catastrophic forgetting of previous APIs.Our method performs comparably to classic CL with full raw data on the MNIST and SVHN in the DFCL-APIs setting. In the DECL-APIs setting, our method achieves 0.97x, 0.75x and 0.69x performance of classic CL on CIFAR10, CIFAR100, and MiniImageNet.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bc4c060c2cb9e928", "title": "Towards Long-Tailed Recognition for Graph Classification via Collaborative Experts", "authors": ["Siyu Yi", "Zhengyang Mao", "Wei Ju", "Yongdao Zhou", "Luchen Liu", "Xiao Luo", "Ming Zhang"], "year": 2023, "published": "2023-08-31T10:12:32Z", "updated": "2023-09-05T14:46:38Z", "arxiv_id": "http://arxiv.org/abs/2308.16609v2", "url": "https://arxiv.org/abs/2308.16609v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.IR", "cs.SI"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["ying"], "key_results": ["Graph classification, aiming at learning the graph-level representations for effective class assignments, has received outstanding achievements, which heavily relies on high-quality datasets that have balanced class distribution.", "Comprehensive experiments are performed on seven widely-used benchmark datasets to demonstrate the superiority of our method CoMe over state-of-the-art baselines."], "abstract": "Graph classification, aiming at learning the graph-level representations for effective class assignments, has received outstanding achievements, which heavily relies on high-quality datasets that have balanced class distribution. In fact, most real-world graph data naturally presents a long-tailed form, where the head classes occupy much more samples than the tail classes, it thus is essential to study the graph-level classification over long-tailed data while still remaining largely unexplored. However, most existing long-tailed learning methods in visions fail to jointly optimize the representation learning and classifier training, as well as neglect the mining of the hard-to-classify classes. Directly applying existing methods to graphs may lead to sub-optimal performance, since the model trained on graphs would be more sensitive to the long-tailed distribution due to the complex topological characteristics. Hence, in this paper, we propose a novel long-tailed graph-level classification framework via Collaborative Multi-expert Learning (CoMe) to tackle the problem. To equilibrate the contributions of head and tail classes, we first develop balanced contrastive learning from the view of representation learning, and then design an individual-expert classifier training based on hard class mining. In addition, we execute gated fusion and disentangled knowledge distillation among the multiple experts to promote the collaboration in a multi-expert framework. Comprehensive experiments are performed on seven widely-used benchmark datasets to demonstrate the superiority of our method CoMe over state-of-the-art baselines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-2d1ec00fc6b3870d", "title": "Thesis Distillation: Investigating The Impact of Bias in NLP Models on Hate Speech Detection", "authors": ["Fatma Elsafoury"], "year": 2023, "published": "2023-08-31T08:40:41Z", "updated": "2023-12-05T11:43:44Z", "arxiv_id": "http://arxiv.org/abs/2308.16549v2", "url": "https://arxiv.org/abs/2308.16549v2", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["ying"], "key_results": ["This paper is a summary of the work done in my PhD thesis."], "abstract": "This paper is a summary of the work done in my PhD thesis. Where I investigate the impact of bias in NLP models on the task of hate speech detection from three perspectives: explainability, offensive stereotyping bias, and fairness. Then, I discuss the main takeaways from my thesis and how they can benefit the broader NLP community. Finally, I discuss important future research directions. The findings of my thesis suggest that the bias in NLP models impacts the task of hate speech detection from all three perspectives. And that unless we start incorporating social sciences in studying bias in NLP models, we will not effectively overcome the current limitations of measuring and mitigating bias in NLP models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-aea820c8a7d1522e", "title": "Adversarial Finetuning with Latent Representation Constraint to Mitigate Accuracy-Robustness Tradeoff", "authors": ["Satoshi Suzuki", "Shin'ya Yamaguchi", "Shoichiro Takeda", "Sekitoshi Kanai", "Naoki Makishima", "Atsushi Ando", "Ryo Masumura"], "year": 2023, "published": "2023-08-31T04:46:12Z", "updated": "2023-08-31T04:46:12Z", "arxiv_id": "http://arxiv.org/abs/2308.16454v1", "url": "https://arxiv.org/abs/2308.16454v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["yielding"], "key_results": ["This paper addresses the tradeoff between standard accuracy on clean examples and robustness against adversarial examples in deep neural networks (DNNs).", "Although adversarial training (AT) improves robustness, it degrades the standard accuracy, thus yielding the tradeoff.", "By combining these components, ARREST achieves both high standard accuracy and robustness."], "abstract": "This paper addresses the tradeoff between standard accuracy on clean examples and robustness against adversarial examples in deep neural networks (DNNs). Although adversarial training (AT) improves robustness, it degrades the standard accuracy, thus yielding the tradeoff. To mitigate this tradeoff, we propose a novel AT method called ARREST, which comprises three components: (i) adversarial finetuning (AFT), (ii) representation-guided knowledge distillation (RGKD), and (iii) noisy replay (NR). AFT trains a DNN on adversarial examples by initializing its parameters with a DNN that is standardly pretrained on clean examples. RGKD and NR respectively entail a regularization term and an algorithm to preserve latent representations of clean examples during AFT. RGKD penalizes the distance between the representations of the standardly pretrained and AFT DNNs. NR switches input adversarial examples to nonadversarial ones when the representation changes significantly during AFT. By combining these components, ARREST achieves both high standard accuracy and robustness. Experimental results demonstrate that ARREST mitigates the tradeoff more effectively than previous AT-based methods do.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-007299867c32ce35", "title": "Knowledge Distillation from Non-streaming to Streaming ASR Encoder using Auxiliary Non-streaming Layer", "authors": ["Kyuhong Shim", "Jinkyu Lee", "Simyung Chang", "Kyuwoong Hwang"], "year": 2023, "published": "2023-08-31T02:58:33Z", "updated": "2023-08-31T02:58:33Z", "arxiv_id": "http://arxiv.org/abs/2308.16415v1", "url": "https://arxiv.org/abs/2308.16415v1", "primary_category": "cs.CL", "categories": ["cs.CL", "eess.AS"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["To improve the performance of streaming ASR, knowledge distillation (KD) from the non-streaming to streaming model has been studied, mainly focusing on aligning the output token probabilities."], "abstract": "Streaming automatic speech recognition (ASR) models are restricted from accessing future context, which results in worse performance compared to the non-streaming models. To improve the performance of streaming ASR, knowledge distillation (KD) from the non-streaming to streaming model has been studied, mainly focusing on aligning the output token probabilities. In this paper, we propose a layer-to-layer KD from the teacher encoder to the student encoder. To ensure that features are extracted using the same context, we insert auxiliary non-streaming branches to the student and perform KD from the non-streaming teacher layer to the non-streaming auxiliary layer. We design a special KD loss that leverages the autoregressive predictive coding (APC) mechanism to encourage the streaming model to predict unseen future contexts. Experimental results show that the proposed method can significantly reduce the word error rate compared to previous token probability distillation methods.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-38450be896e3d493", "title": "LM-Infinite: Zero-Shot Extreme Length Generalization for Large Language Models", "authors": ["Chi Han", "Qifan Wang", "Hao Peng", "Wenhan Xiong", "Yu Chen", "Heng Ji", "Sinong Wang"], "year": 2023, "published": "2023-08-30T16:47:51Z", "updated": "2024-06-24T21:22:00Z", "arxiv_id": "http://arxiv.org/abs/2308.16137v7", "url": "https://arxiv.org/abs/2308.16137v7", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Long Context", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["It also improves performance on downstream tasks such as Passkey Retrieval and Qasper in the zero-shot setting.", "LM-Infinite brings substantial efficiency improvements: it achieves 2.7x decoding speed up and 7.5x memory saving over the original model."], "abstract": "Today's large language models (LLMs) typically train on short text segments (e.g., <4K tokens) due to the quadratic complexity of their Transformer architectures. As a result, their performance suffers drastically on inputs longer than those encountered during training, substantially limiting their applications in real-world tasks involving long contexts such as encoding scientific articles, code repositories, or long dialogues. Through theoretical analysis and empirical investigation, this work identifies three major factors contributing to this length generalization failure. Our theoretical analysis further reveals that commonly used techniques like truncating the attention window or relative positional encodings are inadequate to address them. Answering these challenges, we propose LM-Infinite, a simple and effective method for enhancing LLMs' capabilities of handling long contexts. LM-Infinite is highly flexible and can be used with most modern LLMs off-the-shelf. Without any parameter updates, it allows LLMs pre-trained with 2K or 4K-long segments to generalize to up to 200M length inputs while retaining perplexity. It also improves performance on downstream tasks such as Passkey Retrieval and Qasper in the zero-shot setting. LM-Infinite brings substantial efficiency improvements: it achieves 2.7x decoding speed up and 7.5x memory saving over the original model. Our codes are released at \\url{https://github.com/Glaciohound/LM-Infinite}.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6fa1a413a6eb01a4", "title": "A Parameter-Free Two-Bit Covariance Estimator with Improved Operator Norm Error Rate", "authors": ["Junren Chen", "Michael K. Ng"], "year": 2023, "published": "2023-08-30T14:31:24Z", "updated": "2024-11-10T12:10:54Z", "arxiv_id": "http://arxiv.org/abs/2308.16059v2", "url": "https://arxiv.org/abs/2308.16059v2", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.IT", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["ying"], "key_results": ["The estimator achieves near minimax rate for general sub-Gaussian distributions, but also suffers from two downsides: theoretically, there is an essential gap on operator norm error between their estimator and sample covariance when the diagonal of the covariance matrix is dominated by only a few entries; practically, its performance heavily relies on the dithering scale, which needs to be tuned according to some unknown parameters.", "By employing dithering scales varying across entries, our estimator enjoys an improved operator norm error rate that depends on the effective rank of the underlying covariance matrix rather than the ambient dimension, thus closing the theoretical gap.", "Remarkably, by halving the dithering scales, our estimator oftentimes achieves operator norm errors less than twice of the errors of sample covariance."], "abstract": "A covariance matrix estimator using two bits per entry was recently developed by Dirksen, Maly and Rauhut [Annals of Statistics, 50(6), pp. 3538-3562]. The estimator achieves near minimax rate for general sub-Gaussian distributions, but also suffers from two downsides: theoretically, there is an essential gap on operator norm error between their estimator and sample covariance when the diagonal of the covariance matrix is dominated by only a few entries; practically, its performance heavily relies on the dithering scale, which needs to be tuned according to some unknown parameters. In this work, we propose a new 2-bit covariance matrix estimator that simultaneously addresses both issues. Unlike the sign quantizer associated with uniform dither in Dirksen et al., we adopt a triangular dither prior to a 2-bit quantizer inspired by the multi-bit uniform quantizer. By employing dithering scales varying across entries, our estimator enjoys an improved operator norm error rate that depends on the effective rank of the underlying covariance matrix rather than the ambient dimension, thus closing the theoretical gap. Moreover, our proposed method eliminates the need of any tuning parameter, as the dithering scales are entirely determined by the data. Experimental results under Gaussian samples are provided to showcase the impressive numerical performance of our estimator. Remarkably, by halving the dithering scales, our estimator oftentimes achieves operator norm errors less than twice of the errors of sample covariance.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7356d61289ab4313", "title": "FPTQ: Fine-grained Post-Training Quantization for Large Language Models", "authors": ["Qingyuan Li", "Yifan Zhang", "Liang Li", "Peng Yao", "Bo Zhang", "Xiangxiang Chu", "Yerui Sun", "Li Du", "Yuchen Xie"], "year": 2023, "published": "2023-08-30T12:18:18Z", "updated": "2023-08-30T12:18:18Z", "arxiv_id": "http://arxiv.org/abs/2308.15987v1", "url": "https://arxiv.org/abs/2308.15987v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["LLaMA", "LLaMA-2"], "key_results": ["Without whistles and bells, we eliminate the necessity for further fine-tuning and obtain the state-of-the-art W4A8 quantized performance on BLOOM, LLaMA, and LLaMA-2 on standard benchmarks.", "We confirm that the W4A8 quantization is achievable for the deployment of large language models, fostering their wide-spreading real-world applications."], "abstract": "In the era of large-scale language models, the substantial parameter size poses significant challenges for deployment. Being a prevalent compression technique, quantization has emerged as the mainstream practice to tackle this issue, which is mainly centered on two recipes W8A8 and W4A16 (i.e. weights and activations in such bit widths). In this study, we propose a novel W4A8 post-training quantization method for the available open-sourced LLMs, which combines the advantages of both two recipes. Therefore, we can leverage the benefit in the I/O utilization of 4-bit weight quantization and the acceleration due to 8-bit matrix computation. Nevertheless, the W4A8 faces notorious performance degradation. As a remedy, we involve layerwise activation quantization strategies which feature a novel logarithmic equalization for most intractable layers, and we combine them with fine-grained weight quantization. Without whistles and bells, we eliminate the necessity for further fine-tuning and obtain the state-of-the-art W4A8 quantized performance on BLOOM, LLaMA, and LLaMA-2 on standard benchmarks. We confirm that the W4A8 quantization is achievable for the deployment of large language models, fostering their wide-spreading real-world applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-09a5990b3a50b6c4", "title": "Task-Based MoE for Multitask Multilingual Machine Translation", "authors": ["Hai Pham", "Young Jin Kim", "Subhabrata Mukherjee", "David P. Woodruff", "Barnabas Poczos", "Hany Hassan Awadalla"], "year": 2023, "published": "2023-08-30T05:41:29Z", "updated": "2023-10-24T22:59:26Z", "arxiv_id": "http://arxiv.org/abs/2308.15772v3", "url": "https://arxiv.org/abs/2308.15772v3", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["MoE"], "benchmarks": [], "models": [], "key_results": ["Mixture-of-experts (MoE) architecture has been proven a powerful method for diverse tasks in training deep models in many applications."], "abstract": "Mixture-of-experts (MoE) architecture has been proven a powerful method for diverse tasks in training deep models in many applications. However, current MoE implementations are task agnostic, treating all tokens from different tasks in the same manner. In this work, we instead design a novel method that incorporates task information into MoE models at different granular levels with shared dynamic task-based adapters. Our experiments and analysis show the advantages of our approaches over the dense and canonical MoE models on multi-task multilingual machine translations. With task-specific adapters, our models can additionally generalize to new tasks efficiently.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-fa13f391632c214f", "title": "Hanayo: Harnessing Wave-like Pipeline Parallelism for Enhanced Large Model Training Efficiency", "authors": ["Ziming Liu", "Shenggan Cheng", "Haotian Zhou", "Yang You"], "year": 2023, "published": "2023-08-30T05:03:29Z", "updated": "2023-08-30T05:03:29Z", "arxiv_id": "http://arxiv.org/abs/2308.15762v1", "url": "https://arxiv.org/abs/2308.15762v1", "primary_category": "cs.DC", "categories": ["cs.DC"], "pillar": "Efficient Training & Inference", "methods": ["Memory / RAG"], "benchmarks": [], "models": ["BERT"], "key_results": ["Our evaluation, conducted on four distinct computing clusters and involving both GPT-like and BERT-like architectures with up to 32 GPUs, demonstrates up to a 30.4 \\% increase in throughput compared to the state-of-the-art approach."], "abstract": "Large-scale language models have become increasingly challenging and expensive to train. Among various methods addressing this issue, Pipeline Parallelism has been widely employed to accommodate massive model weights within limited GPU memory. This paper introduces Hanayo, a wave-like pipeline parallelism strategy that boasts a concise structure and practical applicability, alongside a high-performance pipeline execution runtime to tackle the challenges of pipeline strategy implementation. Hanayo mitigates the issues of pipeline bubbles and excessive memory consumption prevalent in existing schemes, without resorting to model duplicates as in Chimera. Our evaluation, conducted on four distinct computing clusters and involving both GPT-like and BERT-like architectures with up to 32 GPUs, demonstrates up to a 30.4 \\% increase in throughput compared to the state-of-the-art approach.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c3db863658d42176", "title": "Optimizing Factual Accuracy in Text Generation through Dynamic Knowledge Selection", "authors": ["Hongjin Qian", "Zhicheng Dou", "Jiejun Tan", "Haonan Chen", "Haoqi Gu", "Ruofei Lai", "Xinyu Zhang", "Zhao Cao", "Ji-Rong Wen"], "year": 2023, "published": "2023-08-30T02:22:40Z", "updated": "2023-08-30T02:22:40Z", "arxiv_id": "http://arxiv.org/abs/2308.15711v1", "url": "https://arxiv.org/abs/2308.15711v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["Experiment results show that DKGen outperforms all baseline models."], "abstract": "Language models (LMs) have revolutionized the way we interact with information, but they often generate nonfactual text, raising concerns about their reliability. Previous methods use external knowledge as references for text generation to enhance factuality but often struggle with the knowledge mix-up(e.g., entity mismatch) of irrelevant references. Besides,as the length of the output text grows, the randomness of sampling can escalate, detrimentally impacting the factual accuracy of the generated text. In this paper, we present DKGen, which divide the text generation process into an iterative process. In each iteration, DKGen takes the input query, the previously generated text and a subset of the reference passages as input to generate short text. During the process, the subset is dynamically selected from the full passage set based on their relevance to the previously generated text and the query, largely eliminating the irrelevant references from input. To further enhance DKGen's ability to correctly use these external knowledge, DKGen distills the relevance order of reference passages to the cross-attention distribution of decoder. We train and evaluate DKGen on a large-scale benchmark dataset. Experiment results show that DKGen outperforms all baseline models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-cfc6769b85c47499", "title": "A General Recipe for Automated Machine Learning in Practice", "authors": ["Hernan Ceferino Vazquez"], "year": 2023, "published": "2023-08-29T21:49:28Z", "updated": "2023-08-29T21:49:28Z", "arxiv_id": "http://arxiv.org/abs/2308.15647v1", "url": "https://arxiv.org/abs/2308.15647v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.SE"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["Automated Machine Learning (AutoML) is an area of research that focuses on developing methods to generate machine learning models automatically."], "abstract": "Automated Machine Learning (AutoML) is an area of research that focuses on developing methods to generate machine learning models automatically. The idea of being able to build machine learning models with very little human intervention represents a great opportunity for the practice of applied machine learning. However, there is very little information on how to design an AutoML system in practice. Most of the research focuses on the problems facing optimization algorithms and leaves out the details of how that would be done in practice. In this paper, we propose a frame of reference for building general AutoML systems. Through a narrative review of the main approaches in the area, our main idea is to distill the fundamental concepts in order to support them in a single design. Finally, we discuss some open problems related to the application of AutoML for future research.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-851eb3d777020c83", "title": "On-Device Learning with Binary Neural Networks", "authors": ["Lorenzo Vorabbi", "Davide Maltoni", "Stefano Santi"], "year": 2023, "published": "2023-08-29T13:48:35Z", "updated": "2023-08-29T13:48:35Z", "arxiv_id": "http://arxiv.org/abs/2308.15308v1", "url": "https://arxiv.org/abs/2308.15308v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Existing Continual Learning (CL) solutions only partially address the constraints on power, memory and computation of the deep learning models when deployed on low-power embedded CPUs."], "abstract": "Existing Continual Learning (CL) solutions only partially address the constraints on power, memory and computation of the deep learning models when deployed on low-power embedded CPUs. In this paper, we propose a CL solution that embraces the recent advancements in CL field and the efficiency of the Binary Neural Networks (BNN), that use 1-bit for weights and activations to efficiently execute deep learning models. We propose a hybrid quantization of CWR* (an effective CL approach) that considers differently forward and backward pass in order to retain more precision during gradient update step and at the same time minimizing the latency overhead. The choice of a binary network as backbone is essential to meet the constraints of low power devices and, to the best of authors' knowledge, this is the first attempt to prove on-device learning with BNN. The experimental validation carried out confirms the validity and the suitability of the proposed method.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-3624fd5eb8ac913c", "title": "SpikeBERT: A Language Spikformer Learned from BERT with Knowledge Distillation", "authors": ["Changze Lv", "Tianlong Li", "Jianhan Xu", "Chenxi Gu", "Zixuan Ling", "Cenyuan Zhang", "Xiaoqing Zheng", "Xuanjing Huang"], "year": 2023, "published": "2023-08-29T08:41:16Z", "updated": "2024-02-21T13:20:21Z", "arxiv_id": "http://arxiv.org/abs/2308.15122v4", "url": "https://arxiv.org/abs/2308.15122v4", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["BERT"], "key_results": ["To this end, we improve a recently-proposed spiking Transformer (i.e., Spikformer) to make it possible to process language tasks and propose a two-stage knowledge distillation method for training it, which combines pre-training by distilling knowledge from BERT with a large collection of unlabelled texts and fine-tuning with task-specific instances via knowledge distillation again from the BERT fine-tuned on the same training examples.", "Through extensive experimentation, we show that the models trained with our method, named SpikeBERT, outperform state-of-the-art SNNs and even achieve comparable results to BERTs on text classification tasks for both English and Chinese with much less energy consumption."], "abstract": "Spiking neural networks (SNNs) offer a promising avenue to implement deep neural networks in a more energy-efficient way. However, the network architectures of existing SNNs for language tasks are still simplistic and relatively shallow, and deep architectures have not been fully explored, resulting in a significant performance gap compared to mainstream transformer-based networks such as BERT. To this end, we improve a recently-proposed spiking Transformer (i.e., Spikformer) to make it possible to process language tasks and propose a two-stage knowledge distillation method for training it, which combines pre-training by distilling knowledge from BERT with a large collection of unlabelled texts and fine-tuning with task-specific instances via knowledge distillation again from the BERT fine-tuned on the same training examples. Through extensive experimentation, we show that the models trained with our method, named SpikeBERT, outperform state-of-the-art SNNs and even achieve comparable results to BERTs on text classification tasks for both English and Chinese with much less energy consumption. Our code is available at https://github.com/Lvchangze/SpikeBERT.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d9bfd1916d56a3b9", "title": "Recursively Summarizing Enables Long-Term Dialogue Memory in Large Language Models", "authors": ["Qingyue Wang", "Yanhe Fu", "Yanan Cao", "Shuai Wang", "Zhiliang Tian", "Liang Ding"], "year": 2023, "published": "2023-08-29T04:59:53Z", "updated": "2025-08-25T14:43:13Z", "arxiv_id": "http://arxiv.org/abs/2308.15022v4", "url": "https://arxiv.org/abs/2308.15022v4", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Long Context", "Memory / RAG"], "benchmarks": [], "models": ["GPT-4"], "key_results": ["Recently, large language models (LLMs), such as GPT-4, stand out remarkable conversational abilities, enabling them to engage in dynamic and contextually relevant dialogues across a wide range of topics."], "abstract": "Recently, large language models (LLMs), such as GPT-4, stand out remarkable conversational abilities, enabling them to engage in dynamic and contextually relevant dialogues across a wide range of topics. However, given a long conversation, these chatbots fail to recall past information and tend to generate inconsistent responses. To address this, we propose to recursively generate summaries/ memory using large language models (LLMs) to enhance long-term memory ability. Specifically, our method first stimulates LLMs to memorize small dialogue contexts and then recursively produce new memory using previous memory and following contexts. Finally, the chatbot can easily generate a highly consistent response with the help of the latest memory. We evaluate our method on both open and closed LLMs, and the experiments on the widely-used public dataset show that our method can generate more consistent responses in a long-context conversation. Also, we show that our strategy could nicely complement both long-context (e.g., 8K and 16K) and retrieval-enhanced LLMs, bringing further long-term dialogue performance. Notably, our method is a potential solution to enable the LLM to model the extremely long context. The code and scripts are released.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9ac52d0ec5dada16", "title": "Uncovering the Hidden Cost of Model Compression", "authors": ["Diganta Misra", "Muawiz Chaudhary", "Agam Goyal", "Bharat Runwal", "Pin Yu Chen"], "year": 2023, "published": "2023-08-29T01:47:49Z", "updated": "2024-03-15T21:04:31Z", "arxiv_id": "http://arxiv.org/abs/2308.14969v3", "url": "https://arxiv.org/abs/2308.14969v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["ying"], "key_results": ["A primary objective in model compression is to develop sparse and/or quantized models capable of matching or even surpassing the performance of their over-parameterized, full-precision counterparts."], "abstract": "In an age dominated by resource-intensive foundation models, the ability to efficiently adapt to downstream tasks is crucial. Visual Prompting (VP), drawing inspiration from the prompting techniques employed in Large Language Models (LLMs), has emerged as a pivotal method for transfer learning in the realm of computer vision. As the importance of efficiency continues to rise, research into model compression has become indispensable in alleviating the computational burdens associated with training and deploying over-parameterized neural networks. A primary objective in model compression is to develop sparse and/or quantized models capable of matching or even surpassing the performance of their over-parameterized, full-precision counterparts. Although previous studies have explored the effects of model compression on transfer learning, its impact on visual prompting-based transfer remains unclear. This study aims to bridge this gap, shedding light on the fact that model compression detrimentally impacts the performance of visual prompting-based transfer, particularly evident in scenarios with low data volume. Furthermore, our findings underscore the adverse influence of sparsity on the calibration of downstream visual-prompted models. However, intriguingly, we also illustrate that such negative effects on calibration are not present when models are compressed via quantization. This empirical investigation underscores the need for a nuanced understanding beyond mere accuracy in sparse and quantized settings, thereby paving the way for further exploration in Visual Prompting techniques tailored for sparse and quantized models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d251c44e71c1bc5c", "title": "Low-bit Quantization for Deep Graph Neural Networks with Smoothness-aware Message Propagation", "authors": ["Shuang Wang", "Bahaeddin Eravci", "Rustam Guliyev", "Hakan Ferhatosmanoglu"], "year": 2023, "published": "2023-08-29T00:25:02Z", "updated": "2023-08-29T00:25:02Z", "arxiv_id": "http://arxiv.org/abs/2308.14949v1", "url": "https://arxiv.org/abs/2308.14949v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["The proposed GNN quantizer learns quantization ranges and reduces the model size with comparable accuracy even under low-bit quantization.", "Significant improvements are demonstrated over state-of-the-art quantization methods and deep GNN approaches in both full-precision and quantized models.", "The proposed quantizer demonstrates superior performance in INT2 configurations across all stages of GNN, achieving a notable level of accuracy."], "abstract": "Graph Neural Network (GNN) training and inference involve significant challenges of scalability with respect to both model sizes and number of layers, resulting in degradation of efficiency and accuracy for large and deep GNNs. We present an end-to-end solution that aims to address these challenges for efficient GNNs in resource constrained environments while avoiding the oversmoothing problem in deep GNNs. We introduce a quantization based approach for all stages of GNNs, from message passing in training to node classification, compressing the model and enabling efficient processing. The proposed GNN quantizer learns quantization ranges and reduces the model size with comparable accuracy even under low-bit quantization. To scale with the number of layers, we devise a message propagation mechanism in training that controls layer-wise changes of similarities between neighboring nodes. This objective is incorporated into a Lagrangian function with constraints and a differential multiplier method is utilized to iteratively find optimal embeddings. This mitigates oversmoothing and suppresses the quantization error to a bound. Significant improvements are demonstrated over state-of-the-art quantization methods and deep GNN approaches in both full-precision and quantized models. The proposed quantizer demonstrates superior performance in INT2 configurations across all stages of GNN, achieving a notable level of accuracy. In contrast, existing quantization approaches fail to generate satisfactory accuracy levels. Finally, the inference with INT2 and INT4 representations exhibits a speedup of 5.11 $\\times$ and 4.70 $\\times$ compared to full precision counterparts, respectively.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5130aef5b15feee0", "title": "Maestro: Uncovering Low-Rank Structures via Trainable Decomposition", "authors": ["Samuel Horvath", "Stefanos Laskaridis", "Shashank Rajput", "Hongyi Wang"], "year": 2023, "published": "2023-08-28T23:08:15Z", "updated": "2024-06-14T17:40:29Z", "arxiv_id": "http://arxiv.org/abs/2308.14929v2", "url": "https://arxiv.org/abs/2308.14929v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Deep Research"], "benchmarks": [], "models": ["yields", "ying"], "key_results": ["While achieving high compression rates, they often incur significant computational overheads at training or lead to non-negligible accuracy penalty."], "abstract": "Deep Neural Networks (DNNs) have been a large driver for AI breakthroughs in recent years. However, these models have been getting increasingly large as they become more accurate and safe. This means that their training becomes increasingly costly and time-consuming and typically yields a single model to fit all targets. Various techniques have been proposed in the literature to mitigate this, including pruning, sparsification, or quantization of model weights and updates. While achieving high compression rates, they often incur significant computational overheads at training or lead to non-negligible accuracy penalty. Alternatively, factorization methods have been leveraged for low-rank compression of DNNs. Similarly, such techniques (e.g., SVD) frequently rely on heavy iterative decompositions of layers and are potentially sub-optimal for non-linear models, such as DNNs. We take a further step in designing efficient low-rank models and propose Maestro, a framework for trainable low-rank layers. Instead of iteratively applying a priori decompositions, the low-rank structure is baked into the training process through LoD, a low-rank ordered decomposition. Not only is this the first time importance ordering via sampling is applied on the decomposed DNN structure, but it also allows selecting ranks at a layer granularity. Our theoretical analysis demonstrates that in special cases LoD recovers the SVD decomposition and PCA. Applied to DNNs, Maestro enables the extraction of lower footprint models that preserve performance. Simultaneously, it enables the graceful trade-off between accuracy-latency for deployment to even more constrained devices without retraining.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-a0475f7a286ea1da", "title": "Emergent Linear Representations in World Models of Self-Supervised Sequence Models", "authors": ["Neel Nanda", "Andrew Lee", "Martin Wattenberg"], "year": 2023, "published": "2023-09-02T13:37:34Z", "updated": "2023-09-07T20:36:48Z", "arxiv_id": "http://arxiv.org/abs/2309.00941v2", "url": "https://arxiv.org/abs/2309.00941v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "World Model"], "benchmarks": [], "models": ["ying"], "key_results": ["How do sequence models represent their decision-making process?"], "abstract": "How do sequence models represent their decision-making process? Prior work suggests that Othello-playing neural network learned nonlinear models of the board state (Li et al., 2023). In this work, we provide evidence of a closely related linear representation of the board. In particular, we show that probing for \"my colour\" vs. \"opponent's colour\" may be a simple yet powerful way to interpret the model's internal state. This precise understanding of the internal representations allows us to control the model's behaviour with simple vector arithmetic. Linear representations enable significant interpretability progress, which we demonstrate with further exploration of how the world model is computed.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7419046a5ca9a35b", "title": "Knowledge Graph Embeddings for Multi-Lingual Structured Representations of Radiology Reports", "authors": ["Tom van Sonsbeek", "Xiantong Zhen", "Marcel Worring"], "year": 2023, "published": "2023-09-02T11:46:41Z", "updated": "2023-09-14T14:25:37Z", "arxiv_id": "http://arxiv.org/abs/2309.00917v2", "url": "https://arxiv.org/abs/2309.00917v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": ["Distillation", "Interpretability"], "benchmarks": [], "models": ["ying", "BERT"], "key_results": ["The resulting graph embedding uncovers the underlying relationships among clinical terms, achieving a representation that is better understandable for clinicians and clinically more accurate, without reliance on large pre-training datasets."], "abstract": "The way we analyse clinical texts has undergone major changes over the last years. The introduction of language models such as BERT led to adaptations for the (bio)medical domain like PubMedBERT and ClinicalBERT. These models rely on large databases of archived medical documents. While performing well in terms of accuracy, both the lack of interpretability and limitations to transfer across languages limit their use in clinical setting. We introduce a novel light-weight graph-based embedding method specifically catering radiology reports. It takes into account the structure and composition of the report, while also connecting medical terms in the report through the multi-lingual SNOMED Clinical Terms knowledge base. The resulting graph embedding uncovers the underlying relationships among clinical terms, achieving a representation that is better understandable for clinicians and clinically more accurate, without reliance on large pre-training datasets. We show the use of this embedding on two tasks namely disease classification of X-ray reports and image classification. For disease classification our model is competitive with its BERT-based counterparts, while being magnitudes smaller in size and training data requirements. For image classification, we show the effectiveness of the graph embedding leveraging cross-modal knowledge transfer and show how this method is usable across different languages.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d44fe5b339fcb852", "title": "BLSP: Bootstrapping Language-Speech Pre-training via Behavior Alignment of Continuation Writing", "authors": ["Chen Wang", "Minpeng Liao", "Zhongqiang Huang", "Jinliang Lu", "Junhong Wu", "Yuchen Liu", "Chengqing Zong", "Jiajun Zhang"], "year": 2023, "published": "2023-09-02T11:46:05Z", "updated": "2024-05-28T14:26:28Z", "arxiv_id": "http://arxiv.org/abs/2309.00916v2", "url": "https://arxiv.org/abs/2309.00916v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.SD", "eess.AS"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["We achieve this by learning a lightweight modality adapter between a frozen speech encoder and an LLM, ensuring that the LLM exhibits the same generation behavior regardless of the modality of input: a speech segment or its transcript."], "abstract": "The emergence of large language models (LLMs) has sparked significant interest in extending their remarkable language capabilities to speech. However, modality alignment between speech and text still remains an open problem. Current solutions can be categorized into two strategies. One is a cascaded approach where outputs (tokens or states) of a separately trained speech recognition system are used as inputs for LLMs, which limits their potential in modeling alignment between speech and text. The other is an end-to-end approach that relies on speech instruction data, which is very difficult to collect in large quantities. In this paper, we address these issues and propose the BLSP approach that Bootstraps Language-Speech Pre-training via behavior alignment of continuation writing. We achieve this by learning a lightweight modality adapter between a frozen speech encoder and an LLM, ensuring that the LLM exhibits the same generation behavior regardless of the modality of input: a speech segment or its transcript. The training process can be divided into two steps. The first step prompts an LLM to generate texts with speech transcripts as prefixes, obtaining text continuations. In the second step, these continuations are used as supervised signals to train the modality adapter in an end-to-end manner. We demonstrate that this straightforward process can extend the capabilities of LLMs to speech, enabling speech recognition, speech translation, spoken language understanding, and speech conversation, even in zero-shot cross-lingual scenarios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d89cca085a90492c", "title": "A novel framework employing deep multi-attention channels network for the autonomous detection of metastasizing cells through fluorescence microscopy", "authors": ["Michail Mamalakis", "Sarah C. Macfarlane", "Scott V. Notley", "Annica K. B Gad", "George Panoutsos"], "year": 2023, "published": "2023-09-02T11:20:10Z", "updated": "2023-09-02T11:20:10Z", "arxiv_id": "http://arxiv.org/abs/2309.00911v1", "url": "https://arxiv.org/abs/2309.00911v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["We also paved the way for a possible spatial micrometre-level biomarker for future development of diagnostic tools against metastasis (spatial distribution of vimentin)."], "abstract": "We developed a transparent computational large-scale imaging-based framework that can distinguish between normal and metastasizing human cells. The method relies on fluorescence microscopy images showing the spatial organization of actin and vimentin filaments in normal and metastasizing single cells, using a combination of multi-attention channels network and global explainable techniques. We test a classification between normal cells (Bj primary fibroblast), and their isogenically matched, transformed and invasive counterpart (BjTertSV40TRasV12). Manual annotation is not trivial to automate due to the intricacy of the biologically relevant features. In this research, we utilized established deep learning networks and our new multi-attention channel architecture. To increase the interpretability of the network - crucial for this application area - we developed an interpretable global explainable approach correlating the weighted geometric mean of the total cell images and their local GradCam scores. The significant results from our analysis unprecedently allowed a more detailed, and biologically relevant understanding of the cytoskeletal changes that accompany oncogenic transformation of normal to invasive and metastasizing cells. We also paved the way for a possible spatial micrometre-level biomarker for future development of diagnostic tools against metastasis (spatial distribution of vimentin).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-392aec0d875950cc", "title": "Discovering Predictive Relational Object Symbols with Symbolic Attentive Layers", "authors": ["Alper Ahmetoglu", "Batuhan Celik", "Erhan Oztop", "Emre Ugur"], "year": 2023, "published": "2023-09-02T10:06:10Z", "updated": "2023-09-02T10:06:10Z", "arxiv_id": "http://arxiv.org/abs/2309.00889v1", "url": "https://arxiv.org/abs/2309.00889v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["We compare the performance of our proposed architecture with state-of-the-art symbol discovery methods in a simulated tabletop environment where the robot needs to discover symbols related to the relative positions of objects to predict the observed effect successfully."], "abstract": "In this paper, we propose and realize a new deep learning architecture for discovering symbolic representations for objects and their relations based on the self-supervised continuous interaction of a manipulator robot with multiple objects on a tabletop environment. The key feature of the model is that it can handle a changing number number of objects naturally and map the object-object relations into symbolic domain explicitly. In the model, we employ a self-attention layer that computes discrete attention weights from object features, which are treated as relational symbols between objects. These relational symbols are then used to aggregate the learned object symbols and predict the effects of executed actions on each object. The result is a pipeline that allows the formation of object symbols and relational symbols from a dataset of object features, actions, and effects in an end-to-end manner. We compare the performance of our proposed architecture with state-of-the-art symbol discovery methods in a simulated tabletop environment where the robot needs to discover symbols related to the relative positions of objects to predict the observed effect successfully. Our experiments show that the proposed architecture performs better than other baselines in effect prediction while forming not only object symbols but also relational symbols. Furthermore, we analyze the learned symbols and relational patterns between objects to learn about how the model interprets the environment. Our analysis shows that the learned symbols relate to the relative positions of objects, object types, and their horizontal alignment on the table, which reflect the regularities in the environment.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0a2d219fab7b7bd1", "title": "Value Kaleidoscope: Engaging AI with Pluralistic Human Values, Rights, and Duties", "authors": ["Taylor Sorensen", "Liwei Jiang", "Jena Hwang", "Sydney Levine", "Valentina Pyatkin", "Peter West", "Nouha Dziri", "Ximing Lu", "Kavel Rao", "Chandra Bhagavatula", "Maarten Sap", "John Tasioulas", "Yejin Choi"], "year": 2023, "published": "2023-09-02T01:24:59Z", "updated": "2024-04-02T16:52:03Z", "arxiv_id": "http://arxiv.org/abs/2309.00779v2", "url": "https://arxiv.org/abs/2309.00779v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["GPT-4", "phic", "philosophical", "ying"], "key_results": ["To improve AI systems to better reflect value pluralism, the first-order challenge is to explore the extent to which AI systems can model pluralistic human values, rights, and duties as well as their interaction."], "abstract": "Human values are crucial to human decision-making. Value pluralism is the view that multiple correct values may be held in tension with one another (e.g., when considering lying to a friend to protect their feelings, how does one balance honesty with friendship?). As statistical learners, AI systems fit to averages by default, washing out these potentially irreducible value conflicts. To improve AI systems to better reflect value pluralism, the first-order challenge is to explore the extent to which AI systems can model pluralistic human values, rights, and duties as well as their interaction.\n  We introduce ValuePrism, a large-scale dataset of 218k values, rights, and duties connected to 31k human-written situations. ValuePrism's contextualized values are generated by GPT-4 and deemed high-quality by human annotators 91% of the time. We conduct a large-scale study with annotators across diverse social and demographic backgrounds to try to understand whose values are represented.\n  With ValuePrism, we build Kaleido, an open, light-weight, and structured language-based multi-task model that generates, explains, and assesses the relevance and valence (i.e., support or oppose) of human values, rights, and duties within a specific context. Humans prefer the sets of values output by our system over the teacher GPT-4, finding them more accurate and with broader coverage. In addition, we demonstrate that Kaleido can help explain variability in human decision-making by outputting contrasting values. Finally, we show that Kaleido's representations transfer to other philosophical frameworks and datasets, confirming the benefit of an explicit, modular, and interpretable approach to value pluralism. We hope that our work will serve as a step to making more explicit the implicit values behind human decision-making and to steering AI systems to make decisions that are more in accordance with them.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0d676b2eac4be29a", "title": "Efficient RLHF: Reducing the Memory Usage of PPO", "authors": ["Michael Santacroce", "Yadong Lu", "Han Yu", "Yuanzhi Li", "Yelong Shen"], "year": 2023, "published": "2023-09-01T22:57:20Z", "updated": "2023-09-01T22:57:20Z", "arxiv_id": "http://arxiv.org/abs/2309.00754v1", "url": "https://arxiv.org/abs/2309.00754v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Using LoRA during PPO reduces its memory usage to be smaller than SFT while improving alignment across four public benchmarks, and 2.", "Hydra-PPO reduces the latency per sample of LoRA-PPO by up to 65% while maintaining its performance."], "abstract": "Reinforcement Learning with Human Feedback (RLHF) has revolutionized language modeling by aligning models with human preferences. However, the RL stage, Proximal Policy Optimization (PPO), requires over 3x the memory of Supervised Fine-Tuning (SFT), making it infeasible to use for most practitioners. To address this issue, we present a comprehensive analysis the memory usage, performance, and training time of memory-savings techniques for PPO. We introduce Hydra-RLHF by first integrating the SFT and Reward models and then dynamically turning LoRA \"off\" during training. Our experiments show: 1. Using LoRA during PPO reduces its memory usage to be smaller than SFT while improving alignment across four public benchmarks, and 2. Hydra-PPO reduces the latency per sample of LoRA-PPO by up to 65% while maintaining its performance. Our results demonstrate that Hydra-PPO is a simple and promising solution for enabling more widespread usage of RLHF.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e66b9ad94b9c7e1a", "title": "Let the Models Respond: Interpreting Language Model Detoxification Through the Lens of Prompt Dependence", "authors": ["Daniel Scalena", "Gabriele Sarti", "Malvina Nissim", "Elisabetta Fersini"], "year": 2023, "published": "2023-09-01T22:26:06Z", "updated": "2023-09-01T22:26:06Z", "arxiv_id": "http://arxiv.org/abs/2309.00751v1", "url": "https://arxiv.org/abs/2309.00751v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Interpretability", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Due to language models' propensity to generate toxic or hateful responses, several techniques were developed to align model generations with users' preferences."], "abstract": "Due to language models' propensity to generate toxic or hateful responses, several techniques were developed to align model generations with users' preferences. Despite the effectiveness of such methods in improving the safety of model interactions, their impact on models' internal processes is still poorly understood. In this work, we apply popular detoxification approaches to several language models and quantify their impact on the resulting models' prompt dependence using feature attribution methods. We evaluate the effectiveness of counter-narrative fine-tuning and compare it with reinforcement learning-driven detoxification, observing differences in prompt reliance between the two methods despite their similar detoxification performances.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b350689b8abf399e", "title": "TExplain: Explaining Learned Visual Features via Pre-trained (Frozen) Language Models", "authors": ["Saeid Asgari Taghanaki", "Aliasghar Khani", "Ali Saheb Pasand", "Amir Khasahmadi", "Aditya Sanghi", "Karl D. D. Willis", "Ali Mahdavi-Amiri"], "year": 2023, "published": "2023-09-01T20:59:46Z", "updated": "2024-05-02T03:28:00Z", "arxiv_id": "http://arxiv.org/abs/2309.00733v4", "url": "https://arxiv.org/abs/2309.00733v4", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": ["ImageNet"], "models": [], "key_results": ["Interpreting the learned features of vision models has posed a longstanding challenge in the field of machine learning."], "abstract": "Interpreting the learned features of vision models has posed a longstanding challenge in the field of machine learning. To address this issue, we propose a novel method that leverages the capabilities of language models to interpret the learned features of pre-trained image classifiers. Our method, called TExplain, tackles this task by training a neural network to establish a connection between the feature space of image classifiers and language models. Then, during inference, our approach generates a vast number of sentences to explain the features learned by the classifier for a given image. These sentences are then used to extract the most frequent words, providing a comprehensive understanding of the learned features and patterns within the classifier. Our method, for the first time, utilizes these frequent words corresponding to a visual representation to provide insights into the decision-making process of the independently trained classifier, enabling the detection of spurious correlations, biases, and a deeper comprehension of its behavior. To validate the effectiveness of our approach, we conduct experiments on diverse datasets, including ImageNet-9L and Waterbirds. The results demonstrate the potential of our method to enhance the interpretability and robustness of image classifiers.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-23a610d63942873f", "title": "Deep learning in medical image registration: introduction and survey", "authors": ["Ahmad Hammoudeh", "Stéphane Dupont"], "year": 2023, "published": "2023-09-01T20:35:00Z", "updated": "2024-01-10T13:01:36Z", "arxiv_id": "http://arxiv.org/abs/2309.00727v2", "url": "https://arxiv.org/abs/2309.00727v2", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Image registration (IR) is a process that deforms images to align them with respect to a reference space, making it easier for medical practitioners to examine various medical images in a standardized reference frame, such as having the same rotation and scale."], "abstract": "Image registration (IR) is a process that deforms images to align them with respect to a reference space, making it easier for medical practitioners to examine various medical images in a standardized reference frame, such as having the same rotation and scale. This document introduces image registration using a simple numeric example. It provides a definition of image registration along with a space-oriented symbolic representation. This review covers various aspects of image transformations, including affine, deformable, invertible, and bidirectional transformations, as well as medical image registration algorithms such as Voxelmorph, Demons, SyN, Iterative Closest Point, and SynthMorph. It also explores atlas-based registration and multistage image registration techniques, including coarse-fine and pyramid approaches. Furthermore, this survey paper discusses medical image registration taxonomies, datasets, evaluation measures, such as correlation-based metrics, segmentation-based metrics, processing time, and model size. It also explores applications in image-guided surgery, motion tracking, and tumor diagnosis. Finally, the document addresses future research directions, including the further development of transformers.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-978ba7cc05c44517", "title": "Reinforcement Learning with Human Feedback for Realistic Traffic Simulation", "authors": ["Yulong Cao", "Boris Ivanovic", "Chaowei Xiao", "Marco Pavone"], "year": 2023, "published": "2023-09-01T19:29:53Z", "updated": "2023-09-01T19:29:53Z", "arxiv_id": "http://arxiv.org/abs/2309.00709v1", "url": "https://arxiv.org/abs/2309.00709v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG", "cs.RO"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["In light of the challenges and costs of real-world testing, autonomous vehicle developers often rely on testing in simulation for the creation of reliable systems."], "abstract": "In light of the challenges and costs of real-world testing, autonomous vehicle developers often rely on testing in simulation for the creation of reliable systems. A key element of effective simulation is the incorporation of realistic traffic models that align with human knowledge, an aspect that has proven challenging due to the need to balance realism and diversity. This works aims to address this by developing a framework that employs reinforcement learning with human preference (RLHF) to enhance the realism of existing traffic models. This study also identifies two main challenges: capturing the nuances of human preferences on realism and the unification of diverse traffic simulation models. To tackle these issues, we propose using human feedback for alignment and employ RLHF due to its sample efficiency. We also introduce the first dataset for realism alignment in traffic modeling to support such research. Our framework, named TrafficRLHF, demonstrates its proficiency in generating realistic traffic scenarios that are well-aligned with human preferences, as corroborated by comprehensive evaluations on the nuScenes dataset.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-df2ec8a4cafa9e5b", "title": "Point-Bind & Point-LLM: Aligning Point Cloud with Multi-modality for 3D Understanding, Generation, and Instruction Following", "authors": ["Ziyu Guo", "Renrui Zhang", "Xiangyang Zhu", "Yiwen Tang", "Xianzheng Ma", "Jiaming Han", "Kexin Chen", "Peng Gao", "Xianzhi Li", "Hongsheng Li", "Pheng-Ann Heng"], "year": 2023, "published": "2023-09-01T17:59:47Z", "updated": "2023-09-01T17:59:47Z", "arxiv_id": "http://arxiv.org/abs/2309.00615v1", "url": "https://arxiv.org/abs/2309.00615v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.CL", "cs.LG", "cs.MM"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["LLaMA"], "key_results": ["We introduce Point-Bind, a 3D multi-modality model aligning point clouds with 2D image, language, audio, and video."], "abstract": "We introduce Point-Bind, a 3D multi-modality model aligning point clouds with 2D image, language, audio, and video. Guided by ImageBind, we construct a joint embedding space between 3D and multi-modalities, enabling many promising applications, e.g., any-to-3D generation, 3D embedding arithmetic, and 3D open-world understanding. On top of this, we further present Point-LLM, the first 3D large language model (LLM) following 3D multi-modal instructions. By parameter-efficient fine-tuning techniques, Point-LLM injects the semantics of Point-Bind into pre-trained LLMs, e.g., LLaMA, which requires no 3D instruction data, but exhibits superior 3D and multi-modal question-answering capacity. We hope our work may cast a light on the community for extending 3D point clouds to multi-modality applications. Code is available at https://github.com/ZiyuGuo99/Point-Bind_Point-LLM.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-05ef67b0773fee9a", "title": "Baseline Defenses for Adversarial Attacks Against Aligned Language Models", "authors": ["Neel Jain", "Avi Schwarzschild", "Yuxin Wen", "Gowthami Somepalli", "John Kirchenbauer", "Ping-yeh Chiang", "Micah Goldblum", "Aniruddha Saha", "Jonas Geiping", "Tom Goldstein"], "year": 2023, "published": "2023-09-01T17:59:44Z", "updated": "2023-09-04T17:47:36Z", "arxiv_id": "http://arxiv.org/abs/2309.00614v2", "url": "https://arxiv.org/abs/2309.00614v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL", "cs.CR"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["We evaluate several baseline defense strategies against leading adversarial attacks on LLMs, discussing the various settings in which each is feasible and effective."], "abstract": "As Large Language Models quickly become ubiquitous, it becomes critical to understand their security vulnerabilities. Recent work shows that text optimizers can produce jailbreaking prompts that bypass moderation and alignment. Drawing from the rich body of work on adversarial machine learning, we approach these attacks with three questions: What threat models are practically useful in this domain? How do baseline defense techniques perform in this new domain? How does LLM security differ from computer vision?\n  We evaluate several baseline defense strategies against leading adversarial attacks on LLMs, discussing the various settings in which each is feasible and effective. Particularly, we look at three types of defenses: detection (perplexity based), input preprocessing (paraphrase and retokenization), and adversarial training. We discuss white-box and gray-box settings and discuss the robustness-performance trade-off for each of the defenses considered. We find that the weakness of existing discrete optimizers for text, combined with the relatively high costs of optimization, makes standard adaptive attacks more challenging for LLMs. Future research will be needed to uncover whether more powerful optimizers can be developed, or whether the strength of filtering and preprocessing defenses is greater in the LLMs domain than it has been in computer vision.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-2537c50f3998390b", "title": "Taken out of context: On measuring situational awareness in LLMs", "authors": ["Lukas Berglund", "Asa Cooper Stickland", "Mikita Balesni", "Max Kaufmann", "Meg Tong", "Tomasz Korbak", "Daniel Kokotajlo", "Owain Evans"], "year": 2023, "published": "2023-09-01T17:27:37Z", "updated": "2023-09-01T17:27:37Z", "arxiv_id": "http://arxiv.org/abs/2309.00667v1", "url": "https://arxiv.org/abs/2309.00667v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": ["GPT-3", "LLaMA-1"], "key_results": ["An LLM could exploit situational awareness to achieve a high score on safety tests, while taking harmful actions after deployment.", "For both GPT-3 and LLaMA-1, performance improves with model size."], "abstract": "We aim to better understand the emergence of `situational awareness' in large language models (LLMs). A model is situationally aware if it's aware that it's a model and can recognize whether it's currently in testing or deployment. Today's LLMs are tested for safety and alignment before they are deployed. An LLM could exploit situational awareness to achieve a high score on safety tests, while taking harmful actions after deployment. Situational awareness may emerge unexpectedly as a byproduct of model scaling. One way to better foresee this emergence is to run scaling experiments on abilities necessary for situational awareness. As such an ability, we propose `out-of-context reasoning' (in contrast to in-context learning). We study out-of-context reasoning experimentally. First, we finetune an LLM on a description of a test while providing no examples or demonstrations. At test time, we assess whether the model can pass the test. To our surprise, we find that LLMs succeed on this out-of-context reasoning task. Their success is sensitive to the training setup and only works when we apply data augmentation. For both GPT-3 and LLaMA-1, performance improves with model size. These findings offer a foundation for further empirical study, towards predicting and potentially controlling the emergence of situational awareness in LLMs. Code is available at: https://github.com/AsaCooperStickland/situational-awareness-evals.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0885506bc8219928", "title": "Interpretation of High-Dimensional Linear Regression: Effects of Nullspace and Regularization Demonstrated on Battery Data", "authors": ["Joachim Schaeffer", "Eric Lenz", "William C. Chueh", "Martin Z. Bazant", "Rolf Findeisen", "Richard D. Braatz"], "year": 2023, "published": "2023-09-01T16:20:04Z", "updated": "2023-09-06T17:35:10Z", "arxiv_id": "http://arxiv.org/abs/2309.00564v2", "url": "https://arxiv.org/abs/2309.00564v2", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG", "stat.AP", "stat.ME"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying", "yield"], "key_results": ["Furthermore, we demonstrate that regression methods that do not produce coefficients orthogonal to the nullspace, such as fused lasso, can improve interpretability."], "abstract": "High-dimensional linear regression is important in many scientific fields. This article considers discrete measured data of underlying smooth latent processes, as is often obtained from chemical or biological systems. Interpretation in high dimensions is challenging because the nullspace and its interplay with regularization shapes regression coefficients. The data's nullspace contains all coefficients that satisfy $\\mathbf{Xw}=\\mathbf{0}$, thus allowing very different coefficients to yield identical predictions. We developed an optimization formulation to compare regression coefficients and coefficients obtained by physical engineering knowledge to understand which part of the coefficient differences are close to the nullspace. This nullspace method is tested on a synthetic example and lithium-ion battery data. The case studies show that regularization and z-scoring are design choices that, if chosen corresponding to prior physical knowledge, lead to interpretable regression results. Otherwise, the combination of the nullspace and regularization hinders interpretability and can make it impossible to obtain regression coefficients close to the true coefficients when there is a true underlying linear model. Furthermore, we demonstrate that regression methods that do not produce coefficients orthogonal to the nullspace, such as fused lasso, can improve interpretability. In conclusion, the insights gained from the nullspace perspective help to make informed design choices for building regression models on high-dimensional data and reasoning about potential underlying linear models, which are important for system optimization and improving scientific understanding.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-a4374160298db71c", "title": "No Train Still Gain. Unleash Mathematical Reasoning of Large Language Models with Monte Carlo Tree Search Guided by Energy Function", "authors": ["Haotian Xu"], "year": 2023, "published": "2023-09-01T13:10:54Z", "updated": "2023-09-12T03:03:00Z", "arxiv_id": "http://arxiv.org/abs/2309.03224v3", "url": "https://arxiv.org/abs/2309.03224v3", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Preference Optimization"], "benchmarks": ["GSM8K"], "models": [], "key_results": ["Through extensive experiments on two mathematical reasoning benchmarks, GSM8k and AQUA-RAT, we demonstrate the exceptional capabilities of our method, which significantly improves the pass@1 metric of the fine-tuned model without requiring additional fine-tuning or reinforcement learning with human feedback alignment."], "abstract": "Large language models (LLMs) demonstrate impressive language understanding and contextual learning abilities, making them suitable for natural language processing (NLP) tasks and complex mathematical reasoning. However, when applied to mathematical reasoning tasks, LLMs often struggle to generate correct reasoning steps and answers despite having high probabilities for the solutions. To overcome this limitation and enhance the mathematical reasoning capabilities of fine-tuned LLMs without additional fine-tuning steps, we propose a method that incorporates Monte Carlo Tree Search (MCTS) and a lightweight energy function to rank decision steps and enable immediate reaction and precise reasoning. Specifically, we re-formulate the fine-tuned LLMs into a Residual-based Energy Model (Residual-EBM) and employ noise contrastive estimation to estimate the energy function's parameters. We then utilize MCTS with the energy function as a path verifier to search the output space and evaluate the reasoning path. Through extensive experiments on two mathematical reasoning benchmarks, GSM8k and AQUA-RAT, we demonstrate the exceptional capabilities of our method, which significantly improves the pass@1 metric of the fine-tuned model without requiring additional fine-tuning or reinforcement learning with human feedback alignment.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-17eda00e56ca17b8", "title": "RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback", "authors": ["Harrison Lee", "Samrat Phatale", "Hassan Mansoor", "Thomas Mesnard", "Johan Ferret", "Kellie Lu", "Colton Bishop", "Ethan Hall", "Victor Carbune", "Abhinav Rastogi", "Sushant Prakash"], "year": 2023, "published": "2023-09-01T05:53:33Z", "updated": "2024-09-03T14:01:54Z", "arxiv_id": "http://arxiv.org/abs/2309.00267v3", "url": "https://arxiv.org/abs/2309.00267v3", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Across the tasks of summarization, helpful dialogue generation, and harmless dialogue generation, we show that RLAIF achieves comparable performance to RLHF.", "Furthermore, we take a step towards \"self-improvement\" by demonstrating that RLAIF can outperform a supervised fine-tuned baseline even when the AI labeler is the same size as the policy, or even the exact same checkpoint as the initial policy.", "Finally, we introduce direct-RLAIF (d-RLAIF) - a technique that circumvents RM training by obtaining rewards directly from an off-the-shelf LLM during RL, which achieves superior performance to canonical RLAIF."], "abstract": "Reinforcement learning from human feedback (RLHF) has proven effective in aligning large language models (LLMs) with human preferences, but gathering high-quality preference labels is expensive. RL from AI Feedback (RLAIF), introduced in Bai et al., offers a promising alternative that trains the reward model (RM) on preferences generated by an off-the-shelf LLM. Across the tasks of summarization, helpful dialogue generation, and harmless dialogue generation, we show that RLAIF achieves comparable performance to RLHF. Furthermore, we take a step towards \"self-improvement\" by demonstrating that RLAIF can outperform a supervised fine-tuned baseline even when the AI labeler is the same size as the policy, or even the exact same checkpoint as the initial policy. Finally, we introduce direct-RLAIF (d-RLAIF) - a technique that circumvents RM training by obtaining rewards directly from an off-the-shelf LLM during RL, which achieves superior performance to canonical RLAIF. Our results suggest that RLAIF can achieve performance on-par with using human feedback, offering a potential solution to the scalability limitations of RLHF.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-69e21b5244bb54fa", "title": "Why do universal adversarial attacks work on large language models?: Geometry might be the answer", "authors": ["Varshini Subhash", "Anna Bialas", "Weiwei Pan", "Finale Doshi-Velez"], "year": 2023, "published": "2023-09-01T05:09:49Z", "updated": "2023-09-01T05:09:49Z", "arxiv_id": "http://arxiv.org/abs/2309.00254v1", "url": "https://arxiv.org/abs/2309.00254v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL", "cs.CR"], "pillar": "RL / Alignment / Safety", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": ["ying"], "key_results": ["Transformer based large language models with emergent capabilities are becoming increasingly ubiquitous in society."], "abstract": "Transformer based large language models with emergent capabilities are becoming increasingly ubiquitous in society. However, the task of understanding and interpreting their internal workings, in the context of adversarial attacks, remains largely unsolved. Gradient-based universal adversarial attacks have been shown to be highly effective on large language models and potentially dangerous due to their input-agnostic nature. This work presents a novel geometric perspective explaining universal adversarial attacks on large language models. By attacking the 117M parameter GPT-2 model, we find evidence indicating that universal adversarial triggers could be embedding vectors which merely approximate the semantic information in their adversarial training region. This hypothesis is supported by white-box model analysis comprising dimensionality reduction and similarity measurement of hidden representations. We believe this new geometric perspective on the underlying mechanism driving universal attacks could help us gain deeper insight into the internal workings and failure modes of LLMs, thus enabling their mitigation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-72ab8fac68e92921", "title": "Interpretable Medical Imagery Diagnosis with Self-Attentive Transformers: A Review of Explainable AI for Health Care", "authors": ["Tin Lai"], "year": 2023, "published": "2023-09-01T05:01:52Z", "updated": "2023-09-01T05:01:52Z", "arxiv_id": "http://arxiv.org/abs/2309.00252v1", "url": "https://arxiv.org/abs/2309.00252v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["Vision Transformers (ViT) have emerged as state-of-the-art computer vision models, benefiting from self-attention modules."], "abstract": "Recent advancements in artificial intelligence (AI) have facilitated its widespread adoption in primary medical services, addressing the demand-supply imbalance in healthcare. Vision Transformers (ViT) have emerged as state-of-the-art computer vision models, benefiting from self-attention modules. However, compared to traditional machine-learning approaches, deep-learning models are complex and are often treated as a \"black box\" that can cause uncertainty regarding how they operate. Explainable Artificial Intelligence (XAI) refers to methods that explain and interpret machine learning models' inner workings and how they come to decisions, which is especially important in the medical domain to guide the healthcare decision-making process. This review summarises recent ViT advancements and interpretative approaches to understanding the decision-making process of ViT, enabling transparency in medical diagnosis applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-860eb403906c982c", "title": "Turn Fake into Real: Adversarial Head Turn Attacks Against Deepfake Detection", "authors": ["Weijie Wang", "Zhengyu Zhao", "Nicu Sebe", "Bruno Lepri"], "year": 2023, "published": "2023-09-03T07:01:34Z", "updated": "2023-09-03T07:01:34Z", "arxiv_id": "http://arxiv.org/abs/2309.01104v1", "url": "https://arxiv.org/abs/2309.01104v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.CR", "cs.LG", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": ["yields"], "key_results": ["Malicious use of deepfakes leads to serious public concerns and reduces people's trust in digital media.", "In this paper, we propose adversarial head turn (AdvHeat), the first attempt at 3D adversarial face views against deepfake detectors, based on face view synthesis from a single-view fake image."], "abstract": "Malicious use of deepfakes leads to serious public concerns and reduces people's trust in digital media. Although effective deepfake detectors have been proposed, they are substantially vulnerable to adversarial attacks. To evaluate the detector's robustness, recent studies have explored various attacks. However, all existing attacks are limited to 2D image perturbations, which are hard to translate into real-world facial changes. In this paper, we propose adversarial head turn (AdvHeat), the first attempt at 3D adversarial face views against deepfake detectors, based on face view synthesis from a single-view fake image. Extensive experiments validate the vulnerability of various detectors to AdvHeat in realistic, black-box scenarios. For example, AdvHeat based on a simple random search yields a high attack success rate of 96.8% with 360 searching steps. When additional query access is allowed, we can further reduce the step budget to 50. Additional analyses demonstrate that AdvHeat is better than conventional attacks on both the cross-detector transferability and robustness to defenses. The adversarial images generated by AdvHeat are also shown to have natural looks. Our code, including that for generating a multi-view dataset consisting of 360 synthetic views for each of 1000 IDs from FaceForensics++, is available at https://github.com/twowwj/AdvHeaT.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-cf2955f2e64888b1", "title": "Zero-Shot Recommendations with Pre-Trained Large Language Models for Multimodal Nudging", "authors": ["Rachel M. Harrison", "Anton Dereventsov", "Anton Bibin"], "year": 2023, "published": "2023-09-02T21:29:53Z", "updated": "2023-10-01T02:57:42Z", "arxiv_id": "http://arxiv.org/abs/2309.01026v2", "url": "https://arxiv.org/abs/2309.01026v2", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.IR", "cs.LG", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["We present a method for zero-shot recommendation of multimodal non-stationary content that leverages recent advancements in the field of generative AI."], "abstract": "We present a method for zero-shot recommendation of multimodal non-stationary content that leverages recent advancements in the field of generative AI. We propose rendering inputs of different modalities as textual descriptions and to utilize pre-trained LLMs to obtain their numerical representations by computing semantic embeddings. Once unified representations of all content items are obtained, the recommendation can be performed by computing an appropriate similarity metric between them without any additional learning. We demonstrate our approach on a synthetic multimodal nudging environment, where the inputs consist of tabular, textual, and visual data.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-300dbe55f85146cf", "title": "Constrained CycleGAN for Effective Generation of Ultrasound Sector Images of Improved Spatial Resolution", "authors": ["Xiaofei Sun", "He Li", "Wei-Ning Lee"], "year": 2023, "published": "2023-09-02T17:32:00Z", "updated": "2023-09-02T17:32:00Z", "arxiv_id": "http://arxiv.org/abs/2309.00995v1", "url": "https://arxiv.org/abs/2309.00995v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV"], "pillar": "Multimodal / World Models", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Sector US images with improved spatial resolutions are favorable for accurate quantitative analysis of large and dynamic organs, such as the heart.", "In vitro phantom results demonstrate that CCycleGAN successfully generates images with improved spatial resolution as well as higher peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) compared with benchmarks."], "abstract": "Objective. A phased or a curvilinear array produces ultrasound (US) images with a sector field of view (FOV), which inherently exhibits spatially-varying image resolution with inferior quality in the far zone and towards the two sides azimuthally. Sector US images with improved spatial resolutions are favorable for accurate quantitative analysis of large and dynamic organs, such as the heart. Therefore, this study aims to translate US images with spatially-varying resolution to ones with less spatially-varying resolution. CycleGAN has been a prominent choice for unpaired medical image translation; however, it neither guarantees structural consistency nor preserves backscattering patterns between input and generated images for unpaired US images. Approach. To circumvent this limitation, we propose a constrained CycleGAN (CCycleGAN), which directly performs US image generation with unpaired images acquired by different ultrasound array probes. In addition to conventional adversarial and cycle-consistency losses of CycleGAN, CCycleGAN introduces an identical loss and a correlation coefficient loss based on intrinsic US backscattered signal properties to constrain structural consistency and backscattering patterns, respectively. Instead of post-processed B-mode images, CCycleGAN uses envelope data directly obtained from beamformed radio-frequency signals without any other non-linear postprocessing. Main Results. In vitro phantom results demonstrate that CCycleGAN successfully generates images with improved spatial resolution as well as higher peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) compared with benchmarks. Significance. CCycleGAN-generated US images of the in vivo human beating heart further facilitate higher quality heart wall motion estimation than benchmarks-generated ones, particularly in deep regions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7b5ffab87eda535c", "title": "Bridge Diffusion Model: Bridge Chinese Text-to-Image Diffusion Model with English Communities", "authors": ["Shanyuan Liu", "Bo Cheng", "Yuhang Ma", "Liebucha Wu", "Ao Ma", "Xiaoyu Wu", "Dawei Leng", "Yuhui Yin"], "year": 2023, "published": "2023-09-02T14:30:56Z", "updated": "2026-03-17T22:07:14Z", "arxiv_id": "http://arxiv.org/abs/2309.00952v2", "url": "https://arxiv.org/abs/2309.00952v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Text-to-Image generation (TTI) technologies are advancing rapidly, especially in the English language communities."], "abstract": "Text-to-Image generation (TTI) technologies are advancing rapidly, especially in the English language communities. However, apart from the user input language barrier problem, English-native TTI models inherently carry biases from their English world centric training data, which creates a dilemma for development of other language-native TTI models. One common choice is to fine-tune the English-native TTI model with translated samples. It falls short of fully addressing the model bias problem. Alternatively, training non-English language native models from scratch can effectively resolve the English world bias, but model trained this way would diverge from the English TTI communities, thus not able to utilize the strides continuously gaining in the English TTI communities any more. To build Chinese TTI model meanwhile keep compatibility with the English TTI communities, we propose a novel model structure referred as \"Bridge Diffusion Model\" (BDM). The proposed BDM employs a backbone-branch network structure to learn the Chinese semantics while keep the latent space compatible with the English-native TTI backbone, in an end-to-end manner. The unique advantages of the proposed BDM are that it's not only adept at generating images that precisely depict Chinese semantics, but also compatible with various English-native TTI plugins, such as different checkpoints, LoRA, ControlNet, Dreambooth, and Textual Inversion, etc. Moreover, BDM can concurrently generate content seamlessly combining both Chinese-native and English-native semantics within a single image, fostering cultural interaction.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bdf14b2eec8fc2db", "title": "RenAIssance: A Survey into AI Text-to-Image Generation in the Era of Large Model", "authors": ["Fengxiang Bie", "Yibo Yang", "Zhongzhu Zhou", "Adam Ghanem", "Minjia Zhang", "Zhewei Yao", "Xiaoxia Wu", "Connor Holmes", "Pareesa Golnari", "David A. Clifton", "Yuxiong He", "Dacheng Tao", "Shuaiwen Leon Song"], "year": 2023, "published": "2023-09-02T03:27:20Z", "updated": "2023-09-02T03:27:20Z", "arxiv_id": "http://arxiv.org/abs/2309.00810v1", "url": "https://arxiv.org/abs/2309.00810v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Deep Research", "Video Generation"], "benchmarks": [], "models": ["yield"], "key_results": ["In the era of large models, scaling up model size and the integration with large language models have further improved the performance of TTI models, resulting the generation result nearly indistinguishable from real-world images, revolutionizing the way we retrieval images.", "Following this we provide a detailed comparison and critique of these methods and offer possible pathways of improvement for future work.", "In the future work, we argue that TTI development could yield impressive productivity improvements for creation, particularly in the context of the AIGC era, and could be extended to more complex tasks such as video generation and 3D generation."], "abstract": "Text-to-image generation (TTI) refers to the usage of models that could process text input and generate high fidelity images based on text descriptions. Text-to-image generation using neural networks could be traced back to the emergence of Generative Adversial Network (GAN), followed by the autoregressive Transformer. Diffusion models are one prominent type of generative model used for the generation of images through the systematic introduction of noises with repeating steps. As an effect of the impressive results of diffusion models on image synthesis, it has been cemented as the major image decoder used by text-to-image models and brought text-to-image generation to the forefront of machine-learning (ML) research. In the era of large models, scaling up model size and the integration with large language models have further improved the performance of TTI models, resulting the generation result nearly indistinguishable from real-world images, revolutionizing the way we retrieval images. Our explorative study has incentivised us to think that there are further ways of scaling text-to-image models with the combination of innovative model architectures and prediction enhancement techniques. We have divided the work of this survey into five main sections wherein we detail the frameworks of major literature in order to delve into the different types of text-to-image generation methods. Following this we provide a detailed comparison and critique of these methods and offer possible pathways of improvement for future work. In the future work, we argue that TTI development could yield impressive productivity improvements for creation, particularly in the context of the AIGC era, and could be extended to more complex tasks such as video generation and 3D generation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-46e9e4bbf1518118", "title": "OpenIns3D: Snap and Lookup for 3D Open-vocabulary Instance Segmentation", "authors": ["Zhening Huang", "Xiaoyang Wu", "Xi Chen", "Hengshuang Zhao", "Lei Zhu", "Joan Lasenby"], "year": 2023, "published": "2023-09-01T17:59:56Z", "updated": "2024-08-12T16:58:33Z", "arxiv_id": "http://arxiv.org/abs/2309.00616v5", "url": "https://arxiv.org/abs/2309.00616v5", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["World Model", "VLM"], "benchmarks": [], "models": [], "key_results": ["This approach, yet simple, achieves state-of-the-art performance across a wide range of 3D open-vocabulary tasks, including recognition, object detection, and instance segmentation, on both indoor and outdoor datasets.", "When integrated with powerful 2D open-world models, it achieves excellent results in scene understanding tasks."], "abstract": "In this work, we introduce OpenIns3D, a new 3D-input-only framework for 3D open-vocabulary scene understanding. The OpenIns3D framework employs a \"Mask-Snap-Lookup\" scheme. The \"Mask\" module learns class-agnostic mask proposals in 3D point clouds, the \"Snap\" module generates synthetic scene-level images at multiple scales and leverages 2D vision-language models to extract interesting objects, and the \"Lookup\" module searches through the outcomes of \"Snap\" to assign category names to the proposed masks. This approach, yet simple, achieves state-of-the-art performance across a wide range of 3D open-vocabulary tasks, including recognition, object detection, and instance segmentation, on both indoor and outdoor datasets. Moreover, OpenIns3D facilitates effortless switching between different 2D detectors without requiring retraining. When integrated with powerful 2D open-world models, it achieves excellent results in scene understanding tasks. Furthermore, when combined with LLM-powered 2D models, OpenIns3D exhibits an impressive capability to comprehend and process highly complex text queries that demand intricate reasoning and real-world knowledge. Project page: https://zheninghuang.github.io/OpenIns3D/", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ef153e1346e0ccc3", "title": "Zero-Shot Video Moment Retrieval from Frozen Vision-Language Models", "authors": ["Dezhao Luo", "Jiabo Huang", "Shaogang Gong", "Hailin Jin", "Yang Liu"], "year": 2023, "published": "2023-09-01T13:06:50Z", "updated": "2023-09-01T13:06:50Z", "arxiv_id": "http://arxiv.org/abs/2309.00661v1", "url": "https://arxiv.org/abs/2309.00661v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["Accurate video moment retrieval (VMR) requires universal visual-textual correlations that can handle unknown vocabulary and unseen scenes."], "abstract": "Accurate video moment retrieval (VMR) requires universal visual-textual correlations that can handle unknown vocabulary and unseen scenes. However, the learned correlations are likely either biased when derived from a limited amount of moment-text data which is hard to scale up because of the prohibitive annotation cost (fully-supervised), or unreliable when only the video-text pairwise relationships are available without fine-grained temporal annotations (weakly-supervised). Recently, the vision-language models (VLM) demonstrate a new transfer learning paradigm to benefit different vision tasks through the universal visual-textual correlations derived from large-scale vision-language pairwise web data, which has also shown benefits to VMR by fine-tuning in the target domains. In this work, we propose a zero-shot method for adapting generalisable visual-textual priors from arbitrary VLM to facilitate moment-text alignment, without the need for accessing the VMR data. To this end, we devise a conditional feature refinement module to generate boundary-aware visual features conditioned on text queries to enable better moment boundary understanding. Additionally, we design a bottom-up proposal generation strategy that mitigates the impact of domain discrepancies and breaks down complex-query retrieval tasks into individual action retrievals, thereby maximizing the benefits of VLM. Extensive experiments conducted on three VMR benchmark datasets demonstrate the notable performance advantages of our zero-shot algorithm, especially in the novel-word and novel-location out-of-distribution setups.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9a2bc8e93e4a42ec", "title": "Learning Speech Representation From Contrastive Token-Acoustic Pretraining", "authors": ["Chunyu Qiang", "Hao Li", "Yixin Tian", "Ruibo Fu", "Tao Wang", "Longbiao Wang", "Jianwu Dang"], "year": 2023, "published": "2023-09-01T12:35:43Z", "updated": "2023-12-18T12:49:49Z", "arxiv_id": "http://arxiv.org/abs/2309.00424v5", "url": "https://arxiv.org/abs/2309.00424v5", "primary_category": "eess.AS", "categories": ["eess.AS", "cs.AI", "cs.CL", "cs.SD"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["The CTAP model is trained on 210k speech and phoneme pairs, achieving minimally-supervised TTS, VC, and ASR."], "abstract": "For fine-grained generation and recognition tasks such as minimally-supervised text-to-speech (TTS), voice conversion (VC), and automatic speech recognition (ASR), the intermediate representations extracted from speech should serve as a \"bridge\" between text and acoustic information, containing information from both modalities. The semantic content is emphasized, while the paralinguistic information such as speaker identity and acoustic details should be de-emphasized. However, existing methods for extracting fine-grained intermediate representations from speech suffer from issues of excessive redundancy and dimension explosion. Contrastive learning is a good method for modeling intermediate representations from two modalities. However, existing contrastive learning methods in the audio field focus on extracting global descriptive information for downstream audio classification tasks, making them unsuitable for TTS, VC, and ASR tasks. To address these issues, we propose a method named \"Contrastive Token-Acoustic Pretraining (CTAP)\", which uses two encoders to bring phoneme and speech into a joint multimodal space, learning how to connect phoneme and speech at the frame level. The CTAP model is trained on 210k speech and phoneme pairs, achieving minimally-supervised TTS, VC, and ASR. The proposed CTAP method offers a promising solution for fine-grained generation and recognition downstream tasks in speech processing. We provide a website with audio samples.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-f741e90c6e244cba", "title": "VideoGen: A Reference-Guided Latent Diffusion Approach for High Definition Text-to-Video Generation", "authors": ["Xin Li", "Wenqing Chu", "Ye Wu", "Weihang Yuan", "Fanglong Liu", "Qi Zhang", "Fu Li", "Haocheng Feng", "Errui Ding", "Jingdong Wang"], "year": 2023, "published": "2023-09-01T11:14:43Z", "updated": "2023-09-07T08:11:01Z", "arxiv_id": "http://arxiv.org/abs/2309.00398v2", "url": "https://arxiv.org/abs/2309.00398v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": ["Stable Diffusion"], "key_results": ["Then, we introduce an efficient cascaded latent diffusion module conditioned on both the reference image and the text prompt, for generating latent video representations, followed by a flow-based temporal upsampling step to improve the temporal resolution.", "The main characterises of our approach include: the reference image generated by the text-to-image model improves the visual fidelity; using it as the condition makes the diffusion model focus more on learning the video dynamics; and the video decoder is trained over unlabeled video data, thus benefiting from high-quality easily-available videos.", "VideoGen sets a new state-of-the-art in text-to-video generation in terms of both qualitative and quantitative evaluation."], "abstract": "In this paper, we present VideoGen, a text-to-video generation approach, which can generate a high-definition video with high frame fidelity and strong temporal consistency using reference-guided latent diffusion. We leverage an off-the-shelf text-to-image generation model, e.g., Stable Diffusion, to generate an image with high content quality from the text prompt, as a reference image to guide video generation. Then, we introduce an efficient cascaded latent diffusion module conditioned on both the reference image and the text prompt, for generating latent video representations, followed by a flow-based temporal upsampling step to improve the temporal resolution. Finally, we map latent video representations into a high-definition video through an enhanced video decoder. During training, we use the first frame of a ground-truth video as the reference image for training the cascaded latent diffusion module. The main characterises of our approach include: the reference image generated by the text-to-image model improves the visual fidelity; using it as the condition makes the diffusion model focus more on learning the video dynamics; and the video decoder is trained over unlabeled video data, thus benefiting from high-quality easily-available videos. VideoGen sets a new state-of-the-art in text-to-video generation in terms of both qualitative and quantitative evaluation. See \\url{https://videogen.github.io/VideoGen/} for more samples.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-1c1bc654d1ddf707", "title": "Long-Term Ad Memorability: Understanding & Generating Memorable Ads", "authors": ["Harini SI", "Somesh Singh", "Yaman K Singla", "Aanisha Bhattacharyya", "Veeky Baths", "Changyou Chen", "Rajiv Ratn Shah", "Balaji Krishnamurthy"], "year": 2023, "published": "2023-09-01T10:27:04Z", "updated": "2024-11-30T19:08:48Z", "arxiv_id": "http://arxiv.org/abs/2309.00378v5", "url": "https://arxiv.org/abs/2309.00378v5", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.CV", "cs.HC"], "pillar": "Multimodal / World Models", "methods": ["Memory / RAG", "Deep Research", "VLM"], "benchmarks": [], "models": [], "key_results": ["Henry achieves state-of-the-art performance across all prominent literature memorability datasets."], "abstract": "Despite the importance of long-term memory in marketing and brand building, until now, there has been no large-scale study on the memorability of ads. All previous memorability studies have been conducted on short-term recall on specific content types like action videos. On the other hand, long-term memorability is crucial for the advertising industry, and ads are almost always highly multimodal. Therefore, we release the first memorability dataset, LAMBDA, consisting of 1749 participants and 2205 ads covering 276 brands. Running statistical tests over different participant subpopulations and ad types, we find many interesting insights into what makes an ad memorable, e.g., fast-moving ads are more memorable than those with slower scenes; people who use ad-blockers remember a lower number of ads than those who don't. Next, we present a model, Henry, to predict the memorability of a content. Henry achieves state-of-the-art performance across all prominent literature memorability datasets. It shows strong generalization performance with better results in 0-shot on unseen datasets. Finally, with the intent of memorable ad generation, we present a scalable method to build a high-quality memorable ad generation model by leveraging automatically annotated data. Our approach, SEED (Self rEwarding mEmorability Modeling), starts with a language model trained on LAMBDA as seed data and progressively trains an LLM to generate more memorable ads. We show that the generated advertisements have 44% higher memorability scores than the original ads. We release this large-scale ad dataset, UltraLAMBDA, consisting of 5 million ads. Our code and the datasets, LAMBDA and UltraLAMBDA, are open-sourced at https://behavior-in-the-wild.github.io/memorability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-7fbf68e825d1be64", "title": "Towards Contrastive Learning in Music Video Domain", "authors": ["Karel Veldkamp", "Mariya Hendriksen", "Zoltán Szlávik", "Alexander Keijser"], "year": 2023, "published": "2023-09-01T09:08:21Z", "updated": "2023-09-01T09:08:21Z", "arxiv_id": "http://arxiv.org/abs/2309.00347v1", "url": "https://arxiv.org/abs/2309.00347v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.CV", "cs.MM", "cs.SD", "eess.AS"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Our results indicate that pre-trained networks without contrastive fine-tuning outperform our contrastive learning approach when evaluated on both tasks."], "abstract": "Contrastive learning is a powerful way of learning multimodal representations across various domains such as image-caption retrieval and audio-visual representation learning. In this work, we investigate if these findings generalize to the domain of music videos. Specifically, we create a dual en-coder for the audio and video modalities and train it using a bidirectional contrastive loss. For the experiments, we use an industry dataset containing 550 000 music videos as well as the public Million Song Dataset, and evaluate the quality of learned representations on the downstream tasks of music tagging and genre classification. Our results indicate that pre-trained networks without contrastive fine-tuning outperform our contrastive learning approach when evaluated on both tasks. To gain a better understanding of the reasons contrastive learning was not successful for music videos, we perform a qualitative analysis of the learned representations, revealing why contrastive learning might have difficulties uniting embeddings from two modalities. Based on these findings, we outline possible directions for future work. To facilitate the reproducibility of our results, we share our code and the pre-trained model.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c613b568c0e7e872", "title": "DiffuGen: Adaptable Approach for Generating Labeled Image Datasets using Stable Diffusion Models", "authors": ["Michael Shenoda", "Edward Kim"], "year": 2023, "published": "2023-09-01T04:42:03Z", "updated": "2023-09-01T04:42:03Z", "arxiv_id": "http://arxiv.org/abs/2309.00248v1", "url": "https://arxiv.org/abs/2309.00248v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": ["Stable Diffusion Models. Generating high-quality labeled image datasets is crucial for training accurate and robust machine learning models in the field of computer vision. However", "stable diffusion models to create labeled image datasets efficiently. By leveraging stable diffusion models"], "key_results": ["Generating high-quality labeled image datasets is crucial for training accurate and robust machine learning models in the field of computer vision."], "abstract": "Generating high-quality labeled image datasets is crucial for training accurate and robust machine learning models in the field of computer vision. However, the process of manually labeling real images is often time-consuming and costly. To address these challenges associated with dataset generation, we introduce \"DiffuGen,\" a simple and adaptable approach that harnesses the power of stable diffusion models to create labeled image datasets efficiently. By leveraging stable diffusion models, our approach not only ensures the quality of generated datasets but also provides a versatile solution for label generation. In this paper, we present the methodology behind DiffuGen, which combines the capabilities of diffusion models with two distinct labeling techniques: unsupervised and supervised. Distinctively, DiffuGen employs prompt templating for adaptable image generation and textual inversion to enhance diffusion model capabilities.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-ff26109122b21516", "title": "Image Hijacks: Adversarial Images can Control Generative Models at Runtime", "authors": ["Luke Bailey", "Euan Ong", "Stuart Russell", "Scott Emmons"], "year": 2023, "published": "2023-09-01T03:53:40Z", "updated": "2024-09-17T19:56:09Z", "arxiv_id": "http://arxiv.org/abs/2309.00236v4", "url": "https://arxiv.org/abs/2309.00236v4", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL", "cs.CR"], "pillar": "Multimodal / World Models", "methods": ["Long Context", "Safety / Jailbreak", "VLM"], "benchmarks": [], "models": ["LLaMA-2", "LLaVA"], "key_results": ["Are foundation models secure against malicious actors?", "We study these attacks against LLaVA, a state-of-the-art VLM based on CLIP and LLaMA-2, and find that all attack types achieve a success rate of over 80%."], "abstract": "Are foundation models secure against malicious actors? In this work, we focus on the image input to a vision-language model (VLM). We discover image hijacks, adversarial images that control the behaviour of VLMs at inference time, and introduce the general Behaviour Matching algorithm for training image hijacks. From this, we derive the Prompt Matching method, allowing us to train hijacks matching the behaviour of an arbitrary user-defined text prompt (e.g. 'the Eiffel Tower is now located in Rome') using a generic, off-the-shelf dataset unrelated to our choice of prompt. We use Behaviour Matching to craft hijacks for four types of attack, forcing VLMs to generate outputs of the adversary's choice, leak information from their context window, override their safety training, and believe false statements. We study these attacks against LLaVA, a state-of-the-art VLM based on CLIP and LLaMA-2, and find that all attack types achieve a success rate of over 80%. Moreover, our attacks are automated and require only small image perturbations.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-c0e4e427d8c7681e", "title": "Towards Addressing the Misalignment of Object Proposal Evaluation for Vision-Language Tasks via Semantic Grounding", "authors": ["Joshua Feinglass", "Yezhou Yang"], "year": 2023, "published": "2023-09-01T02:19:41Z", "updated": "2023-09-01T02:19:41Z", "arxiv_id": "http://arxiv.org/abs/2309.00215v1", "url": "https://arxiv.org/abs/2309.00215v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.CL"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["The performance of object proposals generated for VL tasks is currently evaluated across all available annotations, a protocol that we show is misaligned - higher scores do not necessarily correspond to improved performance on downstream VL tasks.", "To this end, we propose evaluating object proposals against only a subset of available annotations, selected by thresholding an annotation importance score.", "We show that our method is consistent and demonstrates greatly improved alignment with annotations selected by image captioning metrics and human annotation when compared against existing techniques."], "abstract": "Object proposal generation serves as a standard pre-processing step in Vision-Language (VL) tasks (image captioning, visual question answering, etc.). The performance of object proposals generated for VL tasks is currently evaluated across all available annotations, a protocol that we show is misaligned - higher scores do not necessarily correspond to improved performance on downstream VL tasks. Our work serves as a study of this phenomenon and explores the effectiveness of semantic grounding to mitigate its effects. To this end, we propose evaluating object proposals against only a subset of available annotations, selected by thresholding an annotation importance score. Importance of object annotations to VL tasks is quantified by extracting relevant semantic information from text describing the image. We show that our method is consistent and demonstrates greatly improved alignment with annotations selected by image captioning metrics and human annotation when compared against existing techniques. Lastly, we compare current detectors used in the Scene Graph Generation (SGG) benchmark as a use case, which serves as an example of when traditional object proposal evaluation techniques are misaligned.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9354a6dabcb66250", "title": "Diffusion Model with Clustering-based Conditioning for Food Image Generation", "authors": ["Yue Han", "Jiangpeng He", "Mridul Gupta", "Edward J. Delp", "Fengqing Zhu"], "year": 2023, "published": "2023-09-01T01:40:39Z", "updated": "2023-09-01T01:40:39Z", "arxiv_id": "http://arxiv.org/abs/2309.00199v1", "url": "https://arxiv.org/abs/2309.00199v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["The proposed method is evaluated on the Food-101 dataset and shows improved performance when compared with existing image generation works."], "abstract": "Image-based dietary assessment serves as an efficient and accurate solution for recording and analyzing nutrition intake using eating occasion images as input. Deep learning-based techniques are commonly used to perform image analysis such as food classification, segmentation, and portion size estimation, which rely on large amounts of food images with annotations for training. However, such data dependency poses significant barriers to real-world applications, because acquiring a substantial, diverse, and balanced set of food images can be challenging. One potential solution is to use synthetic food images for data augmentation. Although existing work has explored the use of generative adversarial networks (GAN) based structures for generation, the quality of synthetic food images still remains subpar. In addition, while diffusion-based generative models have shown promising results for general image generation tasks, the generation of food images can be challenging due to the substantial intra-class variance. In this paper, we investigate the generation of synthetic food images based on the conditional diffusion model and propose an effective clustering-based training framework, named ClusDiff, for generating high-quality and representative food images. The proposed method is evaluated on the Food-101 dataset and shows improved performance when compared with existing image generation works. We also demonstrate that the synthetic food images generated by ClusDiff can help address the severe class imbalance issue in long-tailed food classification using the VFN-LT dataset.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bcb70dce0ff2d66b", "title": "Distraction-free Embeddings for Robust VQA", "authors": ["Atharvan Dogra", "Deeksha Varshney", "Ashwin Kalyan", "Ameet Deshpande", "Neeraj Kumar"], "year": 2023, "published": "2023-08-31T21:02:25Z", "updated": "2023-08-31T21:02:25Z", "arxiv_id": "http://arxiv.org/abs/2309.00133v1", "url": "https://arxiv.org/abs/2309.00133v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["The generation of effective latent representations and their subsequent refinement to incorporate precise information is an essential prerequisite for Vision-Language Understanding (VLU) tasks such as Video Question Answering (VQA)."], "abstract": "The generation of effective latent representations and their subsequent refinement to incorporate precise information is an essential prerequisite for Vision-Language Understanding (VLU) tasks such as Video Question Answering (VQA). However, most existing methods for VLU focus on sparsely sampling or fine-graining the input information (e.g., sampling a sparse set of frames or text tokens), or adding external knowledge. We present a novel \"DRAX: Distraction Removal and Attended Cross-Alignment\" method to rid our cross-modal representations of distractors in the latent space. We do not exclusively confine the perception of any input information from various modalities but instead use an attention-guided distraction removal method to increase focus on task-relevant information in latent embeddings. DRAX also ensures semantic alignment of embeddings during cross-modal fusions. We evaluate our approach on a challenging benchmark (SUTD-TrafficQA dataset), testing the framework's abilities for feature and event queries, temporal relation understanding, forecasting, hypothesis, and causal analysis through extensive experiments.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-608a311812dd8bf7", "title": "AttrSeg: Open-Vocabulary Semantic Segmentation via Attribute Decomposition-Aggregation", "authors": ["Chaofan Ma", "Yuhuan Yang", "Chen Ju", "Fei Zhang", "Ya Zhang", "Yanfeng Wang"], "year": 2023, "published": "2023-08-31T19:34:09Z", "updated": "2024-01-06T04:10:27Z", "arxiv_id": "http://arxiv.org/abs/2309.00096v2", "url": "https://arxiv.org/abs/2309.00096v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["One hierarchical aggregation architecture is further proposed to achieve multi-level aggregations, leveraging the meticulously designed clustering module."], "abstract": "Open-vocabulary semantic segmentation is a challenging task that requires segmenting novel object categories at inference time. Recent studies have explored vision-language pre-training to handle this task, but suffer from unrealistic assumptions in practical scenarios, i.e., low-quality textual category names. For example, this paradigm assumes that new textual categories will be accurately and completely provided, and exist in lexicons during pre-training. However, exceptions often happen when encountering ambiguity for brief or incomplete names, new words that are not present in the pre-trained lexicons, and difficult-to-describe categories for users. To address these issues, this work proposes a novel attribute decomposition-aggregation framework, AttrSeg, inspired by human cognition in understanding new concepts. Specifically, in the decomposition stage, we decouple class names into diverse attribute descriptions to complement semantic contexts from multiple perspectives. Two attribute construction strategies are designed: using large language models for common categories, and involving manually labeling for human-invented categories. In the aggregation stage, we group diverse attributes into an integrated global description, to form a discriminative classifier that distinguishes the target object from others. One hierarchical aggregation architecture is further proposed to achieve multi-level aggregations, leveraging the meticulously designed clustering module. The final results are obtained by computing the similarity between aggregated attributes and images embeddings. To evaluate the effectiveness, we annotate three types of datasets with attribute descriptions, and conduct extensive experiments and ablation studies. The results show the superior performance of attribute decomposition-aggregation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-168741303f3e0771", "title": "StyleInV: A Temporal Style Modulated Inversion Network for Unconditional Video Generation", "authors": ["Yuhan Wang", "Liming Jiang", "Chen Change Loy"], "year": 2023, "published": "2023-08-31T17:59:33Z", "updated": "2023-08-31T17:59:33Z", "arxiv_id": "http://arxiv.org/abs/2308.16909v1", "url": "https://arxiv.org/abs/2308.16909v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Unconditional video generation is a challenging task that involves synthesizing high-quality videos that are both coherent and of extended duration."], "abstract": "Unconditional video generation is a challenging task that involves synthesizing high-quality videos that are both coherent and of extended duration. To address this challenge, researchers have used pretrained StyleGAN image generators for high-quality frame synthesis and focused on motion generator design. The motion generator is trained in an autoregressive manner using heavy 3D convolutional discriminators to ensure motion coherence during video generation. In this paper, we introduce a novel motion generator design that uses a learning-based inversion network for GAN. The encoder in our method captures rich and smooth priors from encoding images to latents, and given the latent of an initially generated frame as guidance, our method can generate smooth future latent by modulating the inversion encoder temporally. Our method enjoys the advantage of sparse training and naturally constrains the generation space of our motion generator with the inversion network guided by the initial frame, eliminating the need for heavy discriminators. Moreover, our method supports style transfer with simple fine-tuning when the encoder is paired with a pretrained StyleGAN generator. Extensive experiments conducted on various benchmarks demonstrate the superiority of our method in generating long and high-resolution videos with decent single-frame quality and temporal consistency.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-3c6af7c884592a4b", "title": "Learning to Taste: A Multimodal Wine Dataset", "authors": ["Thoranna Bender", "Simon Moe Sørensen", "Alireza Kashani", "K. Eldjarn Hjorleifsson", "Grethe Hyldig", "Søren Hauberg", "Serge Belongie", "Frederik Warburg"], "year": 2023, "published": "2023-08-31T17:58:28Z", "updated": "2024-01-15T14:07:55Z", "arxiv_id": "http://arxiv.org/abs/2308.16900v4", "url": "https://arxiv.org/abs/2308.16900v4", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["We demonstrate that this shared concept embedding space improves upon separate embedding spaces for coarse flavor classification (alcohol percentage, country, grape, price, rating) and aligns with the intricate human perception of flavor."], "abstract": "We present WineSensed, a large multimodal wine dataset for studying the relations between visual perception, language, and flavor. The dataset encompasses 897k images of wine labels and 824k reviews of wines curated from the Vivino platform. It has over 350k unique bottlings, annotated with year, region, rating, alcohol percentage, price, and grape composition. We obtained fine-grained flavor annotations on a subset by conducting a wine-tasting experiment with 256 participants who were asked to rank wines based on their similarity in flavor, resulting in more than 5k pairwise flavor distances. We propose a low-dimensional concept embedding algorithm that combines human experience with automatic machine similarity kernels. We demonstrate that this shared concept embedding space improves upon separate embedding spaces for coarse flavor classification (alcohol percentage, country, grape, price, rating) and aligns with the intricate human perception of flavor.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bcb6536a3c104c63", "title": "TouchStone: Evaluating Vision-Language Models by Language Models", "authors": ["Shuai Bai", "Shusheng Yang", "Jinze Bai", "Peng Wang", "Xingxuan Zhang", "Junyang Lin", "Xinggang Wang", "Chang Zhou", "Jingren Zhou"], "year": 2023, "published": "2023-08-31T17:52:04Z", "updated": "2023-09-04T15:06:15Z", "arxiv_id": "http://arxiv.org/abs/2308.16890v2", "url": "https://arxiv.org/abs/2308.16890v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.CL"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["GPT-4"], "key_results": ["Large vision-language models (LVLMs) have recently witnessed rapid advancements, exhibiting a remarkable capacity for perceiving, understanding, and processing visual information by connecting visual receptor with large language models (LLMs)."], "abstract": "Large vision-language models (LVLMs) have recently witnessed rapid advancements, exhibiting a remarkable capacity for perceiving, understanding, and processing visual information by connecting visual receptor with large language models (LLMs). However, current assessments mainly focus on recognizing and reasoning abilities, lacking direct evaluation of conversational skills and neglecting visual storytelling abilities. In this paper, we propose an evaluation method that uses strong LLMs as judges to comprehensively evaluate the various abilities of LVLMs. Firstly, we construct a comprehensive visual dialogue dataset TouchStone, consisting of open-world images and questions, covering five major categories of abilities and 27 subtasks. This dataset not only covers fundamental recognition and comprehension but also extends to literary creation. Secondly, by integrating detailed image annotations we effectively transform the multimodal input content into a form understandable by LLMs. This enables us to employ advanced LLMs for directly evaluating the quality of the multimodal dialogue without requiring human intervention. Through validation, we demonstrate that powerful LVLMs, such as GPT-4, can effectively score dialogue quality by leveraging their textual capabilities alone, aligning with human preferences. We hope our work can serve as a touchstone for LVLMs' evaluation and pave the way for building stronger LVLMs. The evaluation code is available at https://github.com/OFA-Sys/TouchStone.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-68add20a46e7b048", "title": "Diffusion Models for Interferometric Satellite Aperture Radar", "authors": ["Alexandre Tuel", "Thomas Kerdreux", "Claudia Hulbert", "Bertrand Rouet-Leduc"], "year": 2023, "published": "2023-08-31T16:26:17Z", "updated": "2023-11-28T17:24:29Z", "arxiv_id": "http://arxiv.org/abs/2308.16847v2", "url": "https://arxiv.org/abs/2308.16847v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG", "eess.IV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Probabilistic Diffusion Models (PDMs) have recently emerged as a very promising class of generative models, achieving high performance in natural image generation."], "abstract": "Probabilistic Diffusion Models (PDMs) have recently emerged as a very promising class of generative models, achieving high performance in natural image generation. However, their performance relative to non-natural images, like radar-based satellite data, remains largely unknown. Generating large amounts of synthetic (and especially labelled) satellite data is crucial to implement deep-learning approaches for the processing and analysis of (interferometric) satellite aperture radar data. Here, we leverage PDMs to generate several radar-based satellite image datasets. We show that PDMs succeed in generating images with complex and realistic structures, but that sampling time remains an issue. Indeed, accelerated sampling strategies, which work well on simple image datasets like MNIST, fail on our radar datasets. We provide a simple and versatile open-source https://github.com/thomaskerdreux/PDM_SAR_InSAR_generation to train, sample and evaluate PDMs using any dataset on a single GPU.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-00e194a3e5ff3b5f", "title": "Towards High-Fidelity Text-Guided 3D Face Generation and Manipulation Using only Images", "authors": ["Cuican Yu", "Guansong Lu", "Yihan Zeng", "Jian Sun", "Xiaodan Liang", "Huibin Li", "Zongben Xu", "Songcen Xu", "Wei Zhang", "Hang Xu"], "year": 2023, "published": "2023-08-31T14:26:33Z", "updated": "2023-08-31T14:26:33Z", "arxiv_id": "http://arxiv.org/abs/2308.16758v1", "url": "https://arxiv.org/abs/2308.16758v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Compared to the existing methods, TG-3DFace creates more realistic and aesthetically pleasing 3D faces, boosting 9% multi-view consistency (MVIC) over Latent3D.", "The rendered face images generated by TG-3DFace achieve higher FID and CLIP score than text-to-2D face/image generation models, demonstrating our superiority in generating realistic and semantic-consistent textures."], "abstract": "Generating 3D faces from textual descriptions has a multitude of applications, such as gaming, movie, and robotics. Recent progresses have demonstrated the success of unconditional 3D face generation and text-to-3D shape generation. However, due to the limited text-3D face data pairs, text-driven 3D face generation remains an open problem. In this paper, we propose a text-guided 3D faces generation method, refer as TG-3DFace, for generating realistic 3D faces using text guidance. Specifically, we adopt an unconditional 3D face generation framework and equip it with text conditions, which learns the text-guided 3D face generation with only text-2D face data. On top of that, we propose two text-to-face cross-modal alignment techniques, including the global contrastive learning and the fine-grained alignment module, to facilitate high semantic consistency between generated 3D faces and input texts. Besides, we present directional classifier guidance during the inference process, which encourages creativity for out-of-domain generations. Compared to the existing methods, TG-3DFace creates more realistic and aesthetically pleasing 3D faces, boosting 9% multi-view consistency (MVIC) over Latent3D. The rendered face images generated by TG-3DFace achieve higher FID and CLIP score than text-to-2D face/image generation models, demonstrating our superiority in generating realistic and semantic-consistent textures.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-184358a39e39aa32", "title": "Socratis: Are large multimodal models emotionally aware?", "authors": ["Katherine Deng", "Arijit Ray", "Reuben Tan", "Saadia Gabriel", "Bryan A. Plummer", "Kate Saenko"], "year": 2023, "published": "2023-08-31T13:59:35Z", "updated": "2023-11-02T17:37:53Z", "arxiv_id": "http://arxiv.org/abs/2308.16741v3", "url": "https://arxiv.org/abs/2308.16741v3", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["We benchmark the capability of state-of-the-art multimodal large language models to generate the reasons for feeling an emotion given an IC pair."], "abstract": "Existing emotion prediction benchmarks contain coarse emotion labels which do not consider the diversity of emotions that an image and text can elicit in humans due to various reasons. Learning diverse reactions to multimodal content is important as intelligent machines take a central role in generating and delivering content to society. To address this gap, we propose Socratis, a societal reactions benchmark, where each image-caption (IC) pair is annotated with multiple emotions and the reasons for feeling them. Socratis contains 18K free-form reactions for 980 emotions on 2075 image-caption pairs from 5 widely-read news and image-caption (IC) datasets. We benchmark the capability of state-of-the-art multimodal large language models to generate the reasons for feeling an emotion given an IC pair. Based on a preliminary human study, we observe that humans prefer human-written reasons over 2 times more often than machine-generated ones. This shows our task is harder than standard generation tasks because it starkly contrasts recent findings where humans cannot tell apart machine vs human-written news articles, for instance. We further see that current captioning metrics based on large vision-language models also fail to correlate with human preferences. We hope that these findings and our benchmark will inspire further research on training emotionally aware models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b5f7ce6bb84891f9", "title": "ViLTA: Enhancing Vision-Language Pre-training through Textual Augmentation", "authors": ["Weihan Wang", "Zhen Yang", "Bin Xu", "Juanzi Li", "Yankui Sun"], "year": 2023, "published": "2023-08-31T12:46:36Z", "updated": "2023-08-31T12:46:36Z", "arxiv_id": "http://arxiv.org/abs/2308.16689v1", "url": "https://arxiv.org/abs/2308.16689v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Distillation", "VLM"], "benchmarks": [], "models": [], "key_results": ["Vision-language pre-training (VLP) methods are blossoming recently, and its crucial goal is to jointly learn visual and textual features via a transformer-based architecture, demonstrating promising improvements on a variety of vision-language tasks.", "By leveraging the above techniques, our ViLTA can achieve better performance on various vision-language tasks."], "abstract": "Vision-language pre-training (VLP) methods are blossoming recently, and its crucial goal is to jointly learn visual and textual features via a transformer-based architecture, demonstrating promising improvements on a variety of vision-language tasks. Prior arts usually focus on how to align visual and textual features, but strategies for improving the robustness of model and speeding up model convergence are left insufficiently explored.\n  In this paper, we propose a novel method ViLTA, comprising of two components to further facilitate the model to learn fine-grained representations among image-text pairs. For Masked Language Modeling (MLM), we propose a cross-distillation method to generate soft labels to enhance the robustness of model, which alleviates the problem of treating synonyms of masked words as negative samples in one-hot labels. For Image-Text Matching (ITM), we leverage the current language encoder to synthesize hard negatives based on the context of language input, encouraging the model to learn high-quality representations by increasing the difficulty of the ITM task. By leveraging the above techniques, our ViLTA can achieve better performance on various vision-language tasks. Extensive experiments on benchmark datasets demonstrate that the effectiveness of ViLTA and its promising potential for vision-language pre-training.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-0245978aaad4a9a3", "title": "Generate Your Own Scotland: Satellite Image Generation Conditioned on Maps", "authors": ["Miguel Espinosa", "Elliot J. Crowley"], "year": 2023, "published": "2023-08-31T11:44:40Z", "updated": "2023-08-31T11:44:40Z", "arxiv_id": "http://arxiv.org/abs/2308.16648v1", "url": "https://arxiv.org/abs/2308.16648v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": ["phic", "ying"], "key_results": ["In this paper we show that state-of-the-art pretrained diffusion models can be conditioned on cartographic data to generate realistic satellite images."], "abstract": "Despite recent advancements in image generation, diffusion models still remain largely underexplored in Earth Observation. In this paper we show that state-of-the-art pretrained diffusion models can be conditioned on cartographic data to generate realistic satellite images. We provide two large datasets of paired OpenStreetMap images and satellite views over the region of Mainland Scotland and the Central Belt. We train a ControlNet model and qualitatively evaluate the results, demonstrating that both image quality and map fidelity are possible. Finally, we provide some insights on the opportunities and challenges of applying these models for remote sensing. Our model weights and code for creating the dataset are publicly available at https://github.com/miquel-espinosa/map-sat.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e9f9d60750b6b337", "title": "Detecting Out-of-Context Image-Caption Pairs in News: A Counter-Intuitive Method", "authors": ["Eivind Moholdt", "Sohail Ahmed Khan", "Duc-Tien Dang-Nguyen"], "year": 2023, "published": "2023-08-31T10:16:59Z", "updated": "2023-08-31T10:16:59Z", "arxiv_id": "http://arxiv.org/abs/2308.16611v1", "url": "https://arxiv.org/abs/2308.16611v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["The growth of misinformation and re-contextualized media in social media and news leads to an increasing need for fact-checking methods."], "abstract": "The growth of misinformation and re-contextualized media in social media and news leads to an increasing need for fact-checking methods. Concurrently, the advancement in generative models makes cheapfakes and deepfakes both easier to make and harder to detect. In this paper, we present a novel approach using generative image models to our advantage for detecting Out-of-Context (OOC) use of images-caption pairs in news. We present two new datasets with a total of $6800$ images generated using two different generative models including (1) DALL-E 2, and (2) Stable-Diffusion. We are confident that the method proposed in this paper can further research on generative models in the field of cheapfake detection, and that the resulting datasets can be used to train and evaluate new models aimed at detecting cheapfakes. We run a preliminary qualitative and quantitative analysis to evaluate the performance of each image generation model for this task, and evaluate a handful of methods for computing image similarity.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-fe8fdcccb0812625", "title": "Towards Optimal Patch Size in Vision Transformers for Tumor Segmentation", "authors": ["Ramtin Mojtahedi", "Mohammad Hamghalam", "Richard K. G. Do", "Amber L. Simpson"], "year": 2023, "published": "2023-08-31T09:57:27Z", "updated": "2023-08-31T09:57:27Z", "arxiv_id": "http://arxiv.org/abs/2308.16598v1", "url": "https://arxiv.org/abs/2308.16598v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG"], "pillar": "Multimodal / World Models", "methods": [], "benchmarks": [], "models": [], "key_results": ["While finding an optimal patch size improves the performance of vision transformer-based models on segmentation tasks, it is a time-consuming and challenging procedure.", "Our model showed consistent and improved results when applied to our private multi-resolution mCRC dataset with a smaller average tumor volume."], "abstract": "Detection of tumors in metastatic colorectal cancer (mCRC) plays an essential role in the early diagnosis and treatment of liver cancer. Deep learning models backboned by fully convolutional neural networks (FCNNs) have become the dominant model for segmenting 3D computerized tomography (CT) scans. However, since their convolution layers suffer from limited kernel size, they are not able to capture long-range dependencies and global context. To tackle this restriction, vision transformers have been introduced to solve FCNN's locality of receptive fields. Although transformers can capture long-range features, their segmentation performance decreases with various tumor sizes due to the model sensitivity to the input patch size. While finding an optimal patch size improves the performance of vision transformer-based models on segmentation tasks, it is a time-consuming and challenging procedure. This paper proposes a technique to select the vision transformer's optimal input multi-resolution image patch size based on the average volume size of metastasis lesions. We further validated our suggested framework using a transfer-learning technique, demonstrating that the highest Dice similarity coefficient (DSC) performance was obtained by pre-training on training data with a larger tumour volume using the suggested ideal patch size and then training with a smaller one. We experimentally evaluate this idea through pre-training our model on a multi-resolution public dataset. Our model showed consistent and improved results when applied to our private multi-resolution mCRC dataset with a smaller average tumor volume. This study lays the groundwork for optimizing semantic segmentation of small objects using vision transformers. The implementation source code is available at:https://github.com/Ramtin-Mojtahedi/OVTPS.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d3019cf8da3ac9b0", "title": "Distribution learning via neural differential equations: a nonparametric statistical perspective", "authors": ["Youssef Marzouk", "Zhi Ren", "Sven Wang", "Jakob Zech"], "year": 2023, "published": "2023-09-03T00:21:37Z", "updated": "2023-09-03T00:21:37Z", "arxiv_id": "http://arxiv.org/abs/2309.01043v1", "url": "https://arxiv.org/abs/2309.01043v1", "primary_category": "math.ST", "categories": ["math.ST", "cs.LG", "math.CA", "stat.ML"], "pillar": "Quant × AI", "methods": [], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["While such models have achieved enormous success in machine learning, particularly for generative modeling and density estimation, little is known about their statistical properties."], "abstract": "Ordinary differential equations (ODEs), via their induced flow maps, provide a powerful framework to parameterize invertible transformations for the purpose of representing complex probability distributions. While such models have achieved enormous success in machine learning, particularly for generative modeling and density estimation, little is known about their statistical properties. This work establishes the first general nonparametric statistical convergence analysis for distribution learning via ODE models trained through likelihood maximization. We first prove a convergence theorem applicable to arbitrary velocity field classes $\\mathcal{F}$ satisfying certain simple boundary constraints. This general result captures the trade-off between approximation error (`bias') and the complexity of the ODE model (`variance'). We show that the latter can be quantified via the $C^1$-metric entropy of the class $\\mathcal F$. We then apply this general framework to the setting of $C^k$-smooth target densities, and establish nearly minimax-optimal convergence rates for two relevant velocity field classes $\\mathcal F$: $C^k$ functions and neural networks. The latter is the practically important case of neural ODEs.\n  Our proof techniques require a careful synthesis of (i) analytical stability results for ODEs, (ii) classical theory for sieved M-estimators, and (iii) recent results on approximation rates and metric entropies of neural network classes. The results also provide theoretical insight on how the choice of velocity field class, and the dependence of this choice on sample size $n$ (e.g., the scaling of width, depth, and sparsity of neural network classes), impacts statistical performance.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-31dcf96b8d9c6209", "title": "Network Topology Inference with Sparsity and Laplacian Constraints", "authors": ["Jiaxi Ying", "Xi Han", "Rui Zhou", "Xiwen Wang", "Hing Cheung So"], "year": 2023, "published": "2023-09-02T15:06:30Z", "updated": "2023-09-02T15:06:30Z", "arxiv_id": "http://arxiv.org/abs/2309.00960v1", "url": "https://arxiv.org/abs/2309.00960v1", "primary_category": "cs.LG", "categories": ["cs.LG", "eess.SP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["phical", "ying2020nonconvex"], "key_results": ["We tackle the network topology inference problem by utilizing Laplacian constrained Gaussian graphical models, which recast the task as estimating a precision matrix in the form of a graph Laplacian."], "abstract": "We tackle the network topology inference problem by utilizing Laplacian constrained Gaussian graphical models, which recast the task as estimating a precision matrix in the form of a graph Laplacian. Recent research \\cite{ying2020nonconvex} has uncovered the limitations of the widely used $\\ell_1$-norm in learning sparse graphs under this model: empirically, the number of nonzero entries in the solution grows with the regularization parameter of the $\\ell_1$-norm; theoretically, a large regularization parameter leads to a fully connected (densest) graph. To overcome these challenges, we propose a graph Laplacian estimation method incorporating the $\\ell_0$-norm constraint. An efficient gradient projection algorithm is developed to solve the resulting optimization problem, characterized by sparsity and Laplacian constraints. Through numerical experiments with synthetic and financial time-series datasets, we demonstrate the effectiveness of the proposed method in network topology inference.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-6aace146def9c65a", "title": "DoRA: Domain-Based Self-Supervised Learning Framework for Low-Resource Real Estate Appraisal", "authors": ["Wei-Wei Du", "Wei-Yao Wang", "Wen-Chih Peng"], "year": 2023, "published": "2023-09-02T08:01:32Z", "updated": "2023-09-14T09:24:28Z", "arxiv_id": "http://arxiv.org/abs/2309.00855v3", "url": "https://arxiv.org/abs/2309.00855v3", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["R10", "phic"], "key_results": ["Our benchmark results on three property types of real-world transactions show that DoRA significantly outperforms the SSL baselines for tabular data, the graph-based methods, and the supervised approaches in the few-shot scenarios by at least 7.6% for MAPE, 11.59% for MAE, and 3.34% for HR10%."], "abstract": "The marketplace system connecting demands and supplies has been explored to develop unbiased decision-making in valuing properties. Real estate appraisal serves as one of the high-cost property valuation tasks for financial institutions since it requires domain experts to appraise the estimation based on the corresponding knowledge and the judgment of the market. Existing automated valuation models reducing the subjectivity of domain experts require a large number of transactions for effective evaluation, which is predominantly limited to not only the labeling efforts of transactions but also the generalizability of new developing and rural areas. To learn representations from unlabeled real estate sets, existing self-supervised learning (SSL) for tabular data neglects various important features, and fails to incorporate domain knowledge. In this paper, we propose DoRA, a Domain-based self-supervised learning framework for low-resource Real estate Appraisal. DoRA is pre-trained with an intra-sample geographic prediction as the pretext task based on the metadata of the real estate for equipping the real estate representations with prior domain knowledge. Furthermore, inter-sample contrastive learning is employed to generalize the representations to be robust for limited transactions of downstream tasks. Our benchmark results on three property types of real-world transactions show that DoRA significantly outperforms the SSL baselines for tabular data, the graph-based methods, and the supervised approaches in the few-shot scenarios by at least 7.6% for MAPE, 11.59% for MAE, and 3.34% for HR10%. We expect DoRA to be useful to other financial practitioners with similar marketplace applications who need general models for properties that are newly built and have limited records. The source code is available at https://github.com/wwweiwei/DoRA.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5d8574cb7c17d095", "title": "Rethinking the Power of Graph Canonization in Graph Representation Learning with Stability", "authors": ["Zehao Dong", "Muhan Zhang", "Philip R. O. Payne", "Michael A Province", "Carlos Cruchaga", "Tianyu Zhao", "Fuhai Li", "Yixin Chen"], "year": 2023, "published": "2023-09-01T21:23:04Z", "updated": "2024-02-09T13:34:07Z", "arxiv_id": "http://arxiv.org/abs/2309.00738v3", "url": "https://arxiv.org/abs/2309.00738v3", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": ["phic"], "key_results": ["In graph datasets where the sufficient condition holds, GNNs enhanced by universal graph canonization consistently outperform GNN baselines and successfully improve the SOTA performance up to $31\\%$, providing the optimal solution to numerous challenging real-world graph analytical tasks like gene network representation learning in bioinformatics."], "abstract": "The expressivity of Graph Neural Networks (GNNs) has been studied broadly in recent years to reveal the design principles for more powerful GNNs. Graph canonization is known as a typical approach to distinguish non-isomorphic graphs, yet rarely adopted when developing expressive GNNs. This paper proposes to maximize the expressivity of GNNs by graph canonization, then the power of such GNNs is studies from the perspective of model stability. A stable GNN will map similar graphs to close graph representations in the vectorial space, and the stability of GNNs is critical to generalize their performance to unseen graphs. We theoretically reveal the trade-off of expressivity and stability in graph-canonization-enhanced GNNs. Then we introduce a notion of universal graph canonization as the general solution to address the trade-off and characterize a widely applicable sufficient condition to solve the universal graph canonization. A comprehensive set of experiments demonstrates the effectiveness of the proposed method. In many popular graph benchmark datasets, graph canonization successfully enhances GNNs and provides highly competitive performance, indicating the capability and great potential of proposed method in general graph representation learning. In graph datasets where the sufficient condition holds, GNNs enhanced by universal graph canonization consistently outperform GNN baselines and successfully improve the SOTA performance up to $31\\%$, providing the optimal solution to numerous challenging real-world graph analytical tasks like gene network representation learning in bioinformatics.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5aaee1dbc375dac6", "title": "Concentrated Differential Privacy for Bandits", "authors": ["Achraf Azize", "Debabrota Basu"], "year": 2023, "published": "2023-09-01T16:08:00Z", "updated": "2024-04-15T12:08:53Z", "arxiv_id": "http://arxiv.org/abs/2309.00557v3", "url": "https://arxiv.org/abs/2309.00557v3", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.CR", "cs.IT", "cs.LG", "math.ST"], "pillar": "Quant × AI", "methods": [], "benchmarks": ["MATH"], "models": [], "key_results": ["Bandits serve as the theoretical foundation of sequential learning and an algorithmic foundation of modern recommender systems."], "abstract": "Bandits serve as the theoretical foundation of sequential learning and an algorithmic foundation of modern recommender systems. However, recommender systems often rely on user-sensitive data, making privacy a critical concern. This paper contributes to the understanding of Differential Privacy (DP) in bandits with a trusted centralised decision-maker, and especially the implications of ensuring zero Concentrated Differential Privacy (zCDP). First, we formalise and compare different adaptations of DP to bandits, depending on the considered input and the interaction protocol. Then, we propose three private algorithms, namely AdaC-UCB, AdaC-GOPE and AdaC-OFUL, for three bandit settings, namely finite-armed bandits, linear bandits, and linear contextual bandits. The three algorithms share a generic algorithmic blueprint, i.e. the Gaussian mechanism and adaptive episodes, to ensure a good privacy-utility trade-off. We analyse and upper bound the regret of these three algorithms. Our analysis shows that in all of these settings, the prices of imposing zCDP are (asymptotically) negligible in comparison with the regrets incurred oblivious to privacy. Next, we complement our regret upper bounds with the first minimax lower bounds on the regret of bandits with zCDP. To prove the lower bounds, we elaborate a new proof technique based on couplings and optimal transport. We conclude by experimentally validating our theoretical results for the three different settings of bandits.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-186bcf9dc91e9015", "title": "New metrics for analyzing continual learners", "authors": ["Nicolas Michel", "Giovanni Chierchia", "Romain Negrel", "Jean-François Bercher", "Toshihiko Yamasaki"], "year": 2023, "published": "2023-09-01T13:53:33Z", "updated": "2023-09-01T13:53:33Z", "arxiv_id": "http://arxiv.org/abs/2309.00462v1", "url": "https://arxiv.org/abs/2309.00462v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Through experiments on benchmark datasets, we demonstrate that our proposed metrics can provide new insights into the stability-plasticity trade-off achieved by models in the continual learning environment."], "abstract": "Deep neural networks have shown remarkable performance when trained on independent and identically distributed data from a fixed set of classes. However, in real-world scenarios, it can be desirable to train models on a continuous stream of data where multiple classification tasks are presented sequentially. This scenario, known as Continual Learning (CL) poses challenges to standard learning algorithms which struggle to maintain knowledge of old tasks while learning new ones. This stability-plasticity dilemma remains central to CL and multiple metrics have been proposed to adequately measure stability and plasticity separately. However, none considers the increasing difficulty of the classification task, which inherently results in performance loss for any model. In that sense, we analyze some limitations of current metrics and identify the presence of setup-induced forgetting. Therefore, we propose new metrics that account for the task's increasing difficulty. Through experiments on benchmark datasets, we demonstrate that our proposed metrics can provide new insights into the stability-plasticity trade-off achieved by models in the continual learning environment.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b1cc498254a13d8b", "title": "Chance or Chaos? Fractal geometry aimed to inspect the nature of Bitcoin", "authors": ["Esther Cabezas-Rivas", "Felipe Sánchez-Coll", "Isaac Tormo-Xaixo"], "year": 2023, "published": "2023-09-01T11:06:19Z", "updated": "2023-09-01T11:06:19Z", "arxiv_id": "http://arxiv.org/abs/2309.00390v1", "url": "https://arxiv.org/abs/2309.00390v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["The aim of this paper is to analyse the Bitcoin in order to shed some light on its nature and behaviour."], "abstract": "The aim of this paper is to analyse the Bitcoin in order to shed some light on its nature and behaviour. We select 9 cryptocurrencies that account for almost 75\\% of total market capitalisation and compare their evolution with that of a wide variety of traditional assets: commodities with spot and futures contracts, treasury bonds, stock indices, growth and value stocks. Fractal geometry will be applied to carry out a careful statistical analysis of the performance of the Bitcoin returns. As a main conclusion, we have detected a high degree of persistence in its prices, which decreases the efficiency but increases its predictability. Moreover, we observe that the underlying technology influences price dynamics, with fully decentralised cryptocurrencies being the only ones to exhibit self-similarity features at any time scale.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9cf4020e84669672", "title": "Predicting Financial Market Trends using Time Series Analysis and Natural Language Processing", "authors": ["Ali Asgarov"], "year": 2023, "published": "2023-08-31T21:20:58Z", "updated": "2023-08-31T21:20:58Z", "arxiv_id": "http://arxiv.org/abs/2309.00136v1", "url": "https://arxiv.org/abs/2309.00136v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": ["yielded", "yield"], "key_results": ["Forecasting financial market trends through time series analysis and natural language processing poses a complex and demanding undertaking, owing to the numerous variables that can influence stock prices."], "abstract": "Forecasting financial market trends through time series analysis and natural language processing poses a complex and demanding undertaking, owing to the numerous variables that can influence stock prices. These variables encompass a spectrum of economic and political occurrences, as well as prevailing public attitudes. Recent research has indicated that the expression of public sentiments on social media platforms such as Twitter may have a noteworthy impact on the determination of stock prices. The objective of this study was to assess the viability of Twitter sentiments as a tool for predicting stock prices of major corporations such as Tesla, Apple. Our study has revealed a robust association between the emotions conveyed in tweets and fluctuations in stock prices. Our findings indicate that positivity, negativity, and subjectivity are the primary determinants of fluctuations in stock prices. The data was analyzed utilizing the Long-Short Term Memory neural network (LSTM) model, which is currently recognized as the leading methodology for predicting stock prices by incorporating Twitter sentiments and historical stock prices data. The models utilized in our study demonstrated a high degree of reliability and yielded precise outcomes for the designated corporations. In summary, this research emphasizes the significance of incorporating public opinions into the prediction of stock prices. The application of Time Series Analysis and Natural Language Processing methodologies can yield significant scientific findings regarding financial market patterns, thereby facilitating informed decision-making among investors. The results of our study indicate that the utilization of Twitter sentiments can serve as a potent instrument for forecasting stock prices, and ought to be factored in when formulating investment strategies.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-d4642421bc8a8b28", "title": "Deep Semi-Supervised Anomaly Detection for Finding Fraud in the Futures Market", "authors": ["Timothy DeLise"], "year": 2023, "published": "2023-08-31T19:07:50Z", "updated": "2023-08-31T19:07:50Z", "arxiv_id": "http://arxiv.org/abs/2309.00088v1", "url": "https://arxiv.org/abs/2309.00088v1", "primary_category": "cs.LG", "categories": ["cs.LG", "q-fin.RM"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["We use exclusive proprietary limit order book data from the TMX exchange in Montréal, with a small set of true labeled instances of fraud, to evaluate Deep SAD against its unsupervised predecessor.", "We show that incorporating a small amount of labeled data into an unsupervised anomaly detection framework can greatly improve its accuracy."], "abstract": "Modern financial electronic exchanges are an exciting and fast-paced marketplace where billions of dollars change hands every day. They are also rife with manipulation and fraud. Detecting such activity is a major undertaking, which has historically been a job reserved exclusively for humans. Recently, more research and resources have been focused on automating these processes via machine learning and artificial intelligence. Fraud detection is overwhelmingly associated with the greater field of anomaly detection, which is usually performed via unsupervised learning techniques because of the lack of labeled data needed for supervised learning. However, a small quantity of labeled data does often exist. This research article aims to evaluate the efficacy of a deep semi-supervised anomaly detection technique, called Deep SAD, for detecting fraud in high-frequency financial data. We use exclusive proprietary limit order book data from the TMX exchange in Montréal, with a small set of true labeled instances of fraud, to evaluate Deep SAD against its unsupervised predecessor. We show that incorporating a small amount of labeled data into an unsupervised anomaly detection framework can greatly improve its accuracy.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-cb2ce0374e924d9f", "title": "Linking microblogging sentiments to stock price movement: An application of GPT-4", "authors": ["Rick Steinert", "Saskia Altmann"], "year": 2023, "published": "2023-08-31T14:49:26Z", "updated": "2023-08-31T14:49:26Z", "arxiv_id": "http://arxiv.org/abs/2308.16771v1", "url": "https://arxiv.org/abs/2308.16771v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "q-fin.CP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["GPT-4.", "GPT-4", "ying", "BERT"], "key_results": ["This paper investigates the potential improvement of the GPT-4 Language Learning Model (LLM) in comparison to BERT for modeling same-day daily stock price movements of Apple and Tesla in 2017, based on sentiment analysis of microblogging messages.", "As a result, GPT-4 exhibited substantial accuracy, outperforming BERT in five out of six months and substantially exceeding a naive buy-and-hold strategy, reaching a peak accuracy of 71.47 % in May."], "abstract": "This paper investigates the potential improvement of the GPT-4 Language Learning Model (LLM) in comparison to BERT for modeling same-day daily stock price movements of Apple and Tesla in 2017, based on sentiment analysis of microblogging messages. We recorded daily adjusted closing prices and translated them into up-down movements. Sentiment for each day was extracted from messages on the Stocktwits platform using both LLMs. We develop a novel method to engineer a comprehensive prompt for contextual sentiment analysis which unlocks the true capabilities of modern LLM. This enables us to carefully retrieve sentiments, perceived advantages or disadvantages, and the relevance towards the analyzed company. Logistic regression is used to evaluate whether the extracted message contents reflect stock price movements. As a result, GPT-4 exhibited substantial accuracy, outperforming BERT in five out of six months and substantially exceeding a naive buy-and-hold strategy, reaching a peak accuracy of 71.47 % in May. The study also highlights the importance of prompt engineering in obtaining desired outputs from GPT-4's contextual abilities. However, the costs of deploying GPT-4 and the need for fine-tuning prompts highlight some practical considerations for its use.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5a0574274df51c01", "title": "Dynamic nsNet2: Efficient Deep Noise Suppression with Early Exiting", "authors": ["Riccardo Miccini", "Alaa Zniber", "Clément Laroche", "Tobias Piechowiak", "Martin Schoeberl", "Luca Pezzarossa", "Ouassim Karrakchou", "Jens Sparsø", "Mounir Ghogho"], "year": 2023, "published": "2023-08-31T12:29:24Z", "updated": "2023-08-31T12:29:24Z", "arxiv_id": "http://arxiv.org/abs/2308.16678v1", "url": "https://arxiv.org/abs/2308.16678v1", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.LG", "eess.AS"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Although deep learning has made strides in the field of deep noise suppression, leveraging deep architectures on resource-constrained devices still proved challenging."], "abstract": "Although deep learning has made strides in the field of deep noise suppression, leveraging deep architectures on resource-constrained devices still proved challenging. Therefore, we present an early-exiting model based on nsNet2 that provides several levels of accuracy and resource savings by halting computations at different stages. Moreover, we adapt the original architecture by splitting the information flow to take into account the injected dynamism. We show the trade-offs between performance and computational complexity based on established metrics.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-fb955f42d069cc55", "title": "New general dependence measures: construction, estimation and application to high-frequency stock returns", "authors": ["Aleksy Leeuwenkamp", "Wentao Hu"], "year": 2023, "published": "2023-08-31T12:19:19Z", "updated": "2023-08-31T12:19:19Z", "arxiv_id": "http://arxiv.org/abs/2309.00025v1", "url": "https://arxiv.org/abs/2309.00025v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "econ.EM", "q-fin.PM", "q-fin.RM", "stat.ME"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Thereby we significantly improve on existing (extreme) dependence measures used in asset pricing and statistics."], "abstract": "We propose a set of dependence measures that are non-linear, local, invariant to a wide range of transformations on the marginals, can show tail and risk asymmetries, are always well-defined, are easy to estimate and can be used on any dataset. We propose a nonparametric estimator and prove its consistency and asymptotic normality. Thereby we significantly improve on existing (extreme) dependence measures used in asset pricing and statistics. To show practical utility, we use these measures on high-frequency stock return data around market distress events such as the 2010 Flash Crash and during the GFC. Contrary to ubiquitously used correlations we find that our measures clearly show tail asymmetry, non-linearity, lack of diversification and endogenous buildup of risks present during these distress events. Additionally, our measures anticipate large (joint) losses during the Flash Crash while also anticipating the bounce back and flagging the subsequent market fragility. Our findings have implications for risk management, portfolio construction and hedging at any frequency.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-e1051971cd9d8318", "title": "CktGNN: Circuit Graph Neural Network for Electronic Design Automation", "authors": ["Zehao Dong", "Weidong Cao", "Muhan Zhang", "Dacheng Tao", "Yixin Chen", "Xuan Zhang"], "year": 2023, "published": "2023-08-31T02:20:25Z", "updated": "2024-02-09T13:38:15Z", "arxiv_id": "http://arxiv.org/abs/2308.16406v2", "url": "https://arxiv.org/abs/2308.16406v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["In this way, it significantly improves design efficiency by reducing the number of subgraphs to perform message passing."], "abstract": "The electronic design automation of analog circuits has been a longstanding challenge in the integrated circuit field due to the huge design space and complex design trade-offs among circuit specifications. In the past decades, intensive research efforts have mostly been paid to automate the transistor sizing with a given circuit topology. By recognizing the graph nature of circuits, this paper presents a Circuit Graph Neural Network (CktGNN) that simultaneously automates the circuit topology generation and device sizing based on the encoder-dependent optimization subroutines. Particularly, CktGNN encodes circuit graphs using a two-level GNN framework (of nested GNN) where circuits are represented as combinations of subgraphs in a known subgraph basis. In this way, it significantly improves design efficiency by reducing the number of subgraphs to perform message passing. Nonetheless, another critical roadblock to advancing learning-assisted circuit design automation is a lack of public benchmarks to perform canonical assessment and reproducible research. To tackle the challenge, we introduce Open Circuit Benchmark (OCB), an open-sourced dataset that contains $10$K distinct operational amplifiers with carefully-extracted circuit specifications. OCB is also equipped with communicative circuit generation and evaluation capabilities such that it can help to generalize CktGNN to design various analog circuits by producing corresponding datasets. Experiments on OCB show the extraordinary advantages of CktGNN through representation-based optimization frameworks over other recent powerful GNN baselines and human experts' manual designs. Our work paves the way toward a learning-based open-sourced design automation for analog circuits. Our source code is available at \\url{https://github.com/zehao-dong/CktGNN}.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5c9ca43579d39da7", "title": "Improving the Accuracy of Transaction-Based Ponzi Detection on Ethereum", "authors": ["Phuong Duy Huynh", "Son Hoang Dau", "Xiaodong Li", "Phuc Luong", "Emanuele Viterbo"], "year": 2023, "published": "2023-08-31T01:54:31Z", "updated": "2024-07-18T03:05:50Z", "arxiv_id": "http://arxiv.org/abs/2308.16391v2", "url": "https://arxiv.org/abs/2308.16391v2", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.CE", "cs.LG", "q-fin.ST"], "pillar": "Quant × AI", "methods": ["Deep Research", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["This contract-code-based approach, while achieving very high accuracy, is not robust because a Ponzi developer can fool a detection model by obfuscating the opcode or inventing a new profit distribution logic that cannot be detected.", "On the contrary, a transaction-based approach could improve the robustness of detection because transactions, unlike smart contracts, are harder to be manipulated.", "However, the current transaction-based detection models achieve fairly low accuracy."], "abstract": "The Ponzi scheme, an old-fashioned fraud, is now popular on the Ethereum blockchain, causing considerable financial losses to many crypto investors. A few Ponzi detection methods have been proposed in the literature, most of which detect a Ponzi scheme based on its smart contract source code. This contract-code-based approach, while achieving very high accuracy, is not robust because a Ponzi developer can fool a detection model by obfuscating the opcode or inventing a new profit distribution logic that cannot be detected. On the contrary, a transaction-based approach could improve the robustness of detection because transactions, unlike smart contracts, are harder to be manipulated. However, the current transaction-based detection models achieve fairly low accuracy. In this paper, we aim to improve the accuracy of the transaction-based models by employing time-series features, which turn out to be crucial in capturing the life-time behaviour a Ponzi application but were completely overlooked in previous works. We propose a new set of 85 features (22 known account-based and 63 new time-series features), which allows off-the-shelf machine learning algorithms to achieve up to 30% higher F1-scores compared to existing works.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-b1a2dd4b71f4fc8e", "title": "A new adaptive pricing framework for perpetual protocols using liquidity curves and on-chain oracles", "authors": ["Chester Bella", "Danny Boahen", "Sudeep Biswas"], "year": 2023, "published": "2023-08-30T18:26:51Z", "updated": "2023-08-30T18:26:51Z", "arxiv_id": "http://arxiv.org/abs/2308.16256v1", "url": "https://arxiv.org/abs/2308.16256v1", "primary_category": "q-fin.TR", "categories": ["q-fin.TR"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["This whitepaper introduces an innovative mechanism for pricing perpetual contracts and quoting fees to traders based on current market conditions."], "abstract": "This whitepaper introduces an innovative mechanism for pricing perpetual contracts and quoting fees to traders based on current market conditions. The approach employs liquidity curves and on-chain oracles to establish a new adaptive pricing framework that considers various factors, ensuring pricing stability and predictability. The framework utilizes parabolic and sigmoid functions to quote prices and fees, accounting for liquidity, active long and short positions, and utilization. This whitepaper provides a detailed explanation of how the adaptive pricing framework, in conjunction with liquidity curves, operates through mathematical modeling and compares it to existing solutions. Furthermore, we explore additional features that enhance the overall efficiency of the decentralized perpetual protocol.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-98314849b32f941e", "title": "Vector Autoregression in Cryptocurrency Markets: Unraveling Complex Causal Networks", "authors": ["Cameron Cornell", "Lewis Mitchell", "Matthew Roughan"], "year": 2023, "published": "2023-08-30T05:27:19Z", "updated": "2023-08-30T05:27:19Z", "arxiv_id": "http://arxiv.org/abs/2308.15769v1", "url": "https://arxiv.org/abs/2308.15769v1", "primary_category": "physics.soc-ph", "categories": ["physics.soc-ph", "q-fin.ST"], "pillar": "Quant × AI", "methods": ["Speculative Decoding", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["We demonstrate that simple linear model structure reveals an inherent complexity associated with the interconnected nature of the data, supporting the use of multivariate modelling to prevent surrogate effects and achieve accurate causal representation."], "abstract": "Methodologies to infer financial networks from the price series of speculative assets vary, however, they generally involve bivariate or multivariate predictive modelling to reveal causal and correlational structures within the time series data. The required model complexity intimately relates to the underlying market efficiency, where one expects a highly developed and efficient market to display very few simple relationships in price data. This has spurred research into the applications of complex nonlinear models for developed markets. However, it remains unclear if simple models can provide meaningful and insightful descriptions of the dependency and interconnectedness of the rapidly developed cryptocurrency market. Here we show that multivariate linear models can create informative cryptocurrency networks that reflect economic intuition, and demonstrate the importance of high-influence nodes. The resulting network confirms that node degree, a measure of influence, is significantly correlated to the market capitalisation of each coin ($ρ=0.193$). However, there remains a proportion of nodes whose influence extends beyond what their market capitalisation would imply. We demonstrate that simple linear model structure reveals an inherent complexity associated with the interconnected nature of the data, supporting the use of multivariate modelling to prevent surrogate effects and achieve accurate causal representation. In a reductive experiment we show that most of the network structure is contained within a small portion of the network, consistent with the Pareto principle, whereby a fraction of the inputs generates a large proportion of the effects. Our results demonstrate that simple multivariate models provide nontrivial information about cryptocurrency market dynamics, and that these dynamics largely depend upon a few key high-influence coins.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-bb614a4e127d45c4", "title": "The Financial Market of Environmental Indices", "authors": ["Thisari K. Mahanama", "Abootaleb Shirvani", "Svetlozar Rachev", "Frank J. Fabozzi"], "year": 2023, "published": "2023-08-29T22:50:49Z", "updated": "2023-08-29T22:50:49Z", "arxiv_id": "http://arxiv.org/abs/2308.15661v1", "url": "https://arxiv.org/abs/2308.15661v1", "primary_category": "q-fin.CP", "categories": ["q-fin.CP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["This paper introduces the concept of a global financial market for environmental indices, addressing sustainability concerns and aiming to attract institutional investors."], "abstract": "This paper introduces the concept of a global financial market for environmental indices, addressing sustainability concerns and aiming to attract institutional investors. Risk mitigation measures are implemented to manage inherent risks associated with investments in this new financial market. We monetize the environmental indices using quantitative measures and construct country-specific environmental indices, enabling them to be viewed as dollar-denominated assets. Our primary goal is to encourage the active engagement of institutional investors in portfolio analysis and trading within this emerging financial market. To evaluate and manage investment risks, our approach incorporates financial econometric theory and dynamic asset pricing tools. We provide an econometric analysis that reveals the relationships between environmental and economic indicators in this market. Additionally, we derive financial put options as insurance instruments that can be employed to manage investment risks. Our factor analysis identifies key drivers in the global financial market for environmental indices. To further evaluate the market's performance, we employ pricing options, efficient frontier analysis, and regression analysis. These tools help us assess the efficiency and effectiveness of the market. Overall, our research contributes to the understanding and development of the global financial market for environmental indices.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-9c58d24db3cfd402", "title": "Combining predictive distributions of electricity prices: Does minimizing the CRPS lead to optimal decisions in day-ahead bidding?", "authors": ["Weronika Nitka", "Rafał Weron"], "year": 2023, "published": "2023-08-29T17:10:38Z", "updated": "2023-08-29T17:10:38Z", "arxiv_id": "http://arxiv.org/abs/2308.15443v1", "url": "https://arxiv.org/abs/2308.15443v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "econ.EM", "stat.CO", "stat.ML"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["yield"], "key_results": ["At the same time, methods are being developed to combine predictive distributions, since no model is perfect and averaging generally improves forecasting performance."], "abstract": "Probabilistic price forecasting has recently gained attention in power trading because decisions based on such predictions can yield significantly higher profits than those made with point forecasts alone. At the same time, methods are being developed to combine predictive distributions, since no model is perfect and averaging generally improves forecasting performance. In this article we address the question of whether using CRPS learning, a novel weighting technique minimizing the continuous ranked probability score (CRPS), leads to optimal decisions in day-ahead bidding. To this end, we conduct an empirical study using hourly day-ahead electricity prices from the German EPEX market. We find that increasing the diversity of an ensemble can have a positive impact on accuracy. At the same time, the higher computational cost of using CRPS learning compared to an equal-weighted aggregation of distributions is not offset by higher profits, despite significantly more accurate predictions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-75f05405289c0c7c", "title": "Matbench Discovery -- A framework to evaluate machine learning crystal stability predictions", "authors": ["Janosh Riebesell", "Rhys E. A. Goodall", "Philipp Benner", "Yuan Chiang", "Bowen Deng", "Gerbrand Ceder", "Mark Asta", "Alpha A. Lee", "Anubhav Jain", "Kristin A. Persson"], "year": 2023, "published": "2023-08-28T22:29:57Z", "updated": "2024-12-10T23:52:04Z", "arxiv_id": "http://arxiv.org/abs/2308.14920v3", "url": "https://arxiv.org/abs/2308.14920v3", "primary_category": "cond-mat.mtrl-sci", "categories": ["cond-mat.mtrl-sci", "cs.LG"], "pillar": "Quant × AI", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["yield"], "key_results": ["UIPs emerge as the top performers, achieving F1 scores of 0.57-0.82 and discovery acceleration factors (DAF) of up to 6x on the first 10k stable predictions compared to random selection."], "abstract": "The rapid adoption of machine learning (ML) in domain sciences necessitates best practices and standardized benchmarking for performance evaluation. We present Matbench Discovery, an evaluation framework for ML energy models, applied as pre-filters for high-throughput searches of stable inorganic crystals. This framework addresses the disconnect between thermodynamic stability and formation energy, as well as retrospective vs. prospective benchmarking in materials discovery. We release a Python package to support model submissions and maintain an online leaderboard, offering insights into performance trade-offs. To identify the best-performing ML methodologies for materials discovery, we benchmarked various approaches, including random forests, graph neural networks (GNNs), one-shot predictors, iterative Bayesian optimizers, and universal interatomic potentials (UIP). Our initial results rank models by test set F1 scores for thermodynamic stability prediction: EquiformerV2 + DeNS > Orb > SevenNet > MACE > CHGNet > M3GNet > ALIGNN > MEGNet > CGCNN > CGCNN+P > Wrenformer > BOWSR > Voronoi fingerprint random forest. UIPs emerge as the top performers, achieving F1 scores of 0.57-0.82 and discovery acceleration factors (DAF) of up to 6x on the first 10k stable predictions compared to random selection. We also identify a misalignment between regression metrics and task-relevant classification metrics. Accurate regressors can yield high false-positive rates near the decision boundary at 0 eV/atom above the convex hull. Our results demonstrate UIPs' ability to optimize computational budget allocation for expanding materials databases. However, their limitations remain underexplored in traditional benchmarks. We advocate for task-based evaluation frameworks, as implemented here, to address these limitations and advance ML-guided materials discovery.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-5d775d8be209fb39", "title": "Generating tabular datasets under differential privacy", "authors": ["Gianluca Truda"], "year": 2023, "published": "2023-08-28T16:35:43Z", "updated": "2023-08-28T16:35:43Z", "arxiv_id": "http://arxiv.org/abs/2308.14784v1", "url": "https://arxiv.org/abs/2308.14784v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CR", "cs.DB"], "pillar": "Quant × AI", "methods": ["Video Generation", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Our experiments show that TableDiffusion produces higher-fidelity synthetic datasets, avoids the mode collapse problem, and achieves state-of-the-art performance on privatised tabular data synthesis."], "abstract": "Machine Learning (ML) is accelerating progress across fields and industries, but relies on accessible and high-quality training data. Some of the most important datasets are found in biomedical and financial domains in the form of spreadsheets and relational databases. But this tabular data is often sensitive in nature. Synthetic data generation offers the potential to unlock sensitive data, but generative models tend to memorise and regurgitate training data, which undermines the privacy goal. To remedy this, researchers have incorporated the mathematical framework of Differential Privacy (DP) into the training process of deep neural networks. But this creates a trade-off between the quality and privacy of the resulting data. Generative Adversarial Networks (GANs) are the dominant paradigm for synthesising tabular data under DP, but suffer from unstable adversarial training and mode collapse, which are exacerbated by the privacy constraints and challenging tabular data modality. This work optimises the quality-privacy trade-off of generative models, producing higher quality tabular datasets with the same privacy guarantees. We implement novel end-to-end models that leverage attention mechanisms to learn reversible tabular representations. We also introduce TableDiffusion, the first differentially-private diffusion model for tabular data synthesis. Our experiments show that TableDiffusion produces higher-fidelity synthetic datasets, avoids the mode collapse problem, and achieves state-of-the-art performance on privatised tabular data synthesis. By implementing TableDiffusion to predict the added noise, we enabled it to bypass the challenges of reconstructing mixed-type tabular data. Overall, the diffusion paradigm proves vastly more data and privacy efficient than the adversarial paradigm, due to augmented re-use of each data batch and a smoother iterative training process.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-24d2a8a04f476b48", "title": "Breaking the Bank with ChatGPT: Few-Shot Text Classification for Finance", "authors": ["Lefteris Loukas", "Ilias Stogiannidis", "Prodromos Malakasiotis", "Stavros Vassos"], "year": 2023, "published": "2023-08-28T15:04:16Z", "updated": "2023-08-28T15:04:16Z", "arxiv_id": "http://arxiv.org/abs/2308.14634v1", "url": "https://arxiv.org/abs/2308.14634v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG", "q-fin.CP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["GPT-4", "GPT-3.5", "yielding", "ying"], "key_results": ["Additionally, we fine-tune other pre-trained, masked language models with SetFit, a recent contrastive learning technique, to achieve state-of-the-art results both in full-data and few-shot settings.", "Our findings show that querying GPT-3.5 and GPT-4 can outperform fine-tuned, non-generative models even with fewer examples.", "We conclude that a) our proposed methods offer a practical solution for few-shot tasks in datasets with limited label availability, and b) our state-of-the-art results can inspire future work in the area."], "abstract": "We propose the use of conversational GPT models for easy and quick few-shot text classification in the financial domain using the Banking77 dataset. Our approach involves in-context learning with GPT-3.5 and GPT-4, which minimizes the technical expertise required and eliminates the need for expensive GPU computing while yielding quick and accurate results. Additionally, we fine-tune other pre-trained, masked language models with SetFit, a recent contrastive learning technique, to achieve state-of-the-art results both in full-data and few-shot settings. Our findings show that querying GPT-3.5 and GPT-4 can outperform fine-tuned, non-generative models even with fewer examples. However, subscription fees associated with these solutions may be considered costly for small organizations. Lastly, we find that generative models perform better on the given task when shown representative samples selected by a human expert rather than when shown random ones. We conclude that a) our proposed methods offer a practical solution for few-shot tasks in datasets with limited label availability, and b) our state-of-the-art results can inspire future work in the area.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-daf363e8577e33fc", "title": "An Empirical Analysis on Financial Markets: Insights from the Application of Statistical Physics", "authors": ["Haochen Li", "Yi Cao", "Maria Polukarov", "Carmine Ventre"], "year": 2023, "published": "2023-08-28T00:06:45Z", "updated": "2024-06-25T10:39:56Z", "arxiv_id": "http://arxiv.org/abs/2308.14235v6", "url": "https://arxiv.org/abs/2308.14235v6", "primary_category": "q-fin.TR", "categories": ["q-fin.TR", "q-fin.CP", "q-fin.MF", "q-fin.PR", "q-fin.ST"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["We empirically demonstrate that our model outperforms the benchmarks of traditional approaches and machine learning algorithm."], "abstract": "In this study, we introduce a physical model inspired by statistical physics for predicting price volatility and expected returns by leveraging Level 3 order book data. By drawing parallels between orders in the limit order book and particles in a physical system, we establish unique measures for the system's kinetic energy and momentum as a way to comprehend and evaluate the state of limit order book. Our model goes beyond examining merely the top layers of the order book by introducing the concept of 'active depth', a computationally-efficient approach for identifying order book levels that have impact on price dynamics. We empirically demonstrate that our model outperforms the benchmarks of traditional approaches and machine learning algorithm. Our model provides a nuanced comprehension of market microstructure and produces more accurate forecasts on volatility and expected returns. By incorporating principles of statistical physics, this research offers valuable insights on understanding the behaviours of market participants and order book dynamics.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-2aba8e38ead51f0b", "title": "TimeTrail: Unveiling Financial Fraud Patterns through Temporal Correlation Analysis", "authors": ["Sushrut Ghimire"], "year": 2023, "published": "2023-08-27T22:27:57Z", "updated": "2023-08-27T22:27:57Z", "arxiv_id": "http://arxiv.org/abs/2308.14215v1", "url": "https://arxiv.org/abs/2308.14215v1", "primary_category": "cs.LG", "categories": ["cs.LG", "q-fin.ST"], "pillar": "Quant × AI", "methods": ["Interpretability", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["In the field of financial fraud detection, understanding the underlying patterns and dynamics is important to ensure effective and reliable systems."], "abstract": "In the field of financial fraud detection, understanding the underlying patterns and dynamics is important to ensure effective and reliable systems. This research introduces a new technique, \"TimeTrail,\" which employs advanced temporal correlation analysis to explain complex financial fraud patterns. The technique leverages time-related insights to provide transparent and interpretable explanations for fraud detection decisions, enhancing accountability and trust.\n  The \"TimeTrail\" methodology consists of three key phases: temporal data enrichment, dynamic correlation analysis, and interpretable pattern visualization. Initially, raw financial transaction data is enriched with temporal attributes. Dynamic correlations between these attributes are then quantified using innovative statistical measures. Finally, a unified visualization framework presents these correlations in an interpretable manner. To validate the effectiveness of \"TimeTrail,\" a study is conducted on a diverse financial dataset, surrounding various fraud scenarios. Results demonstrate the technique's capability to uncover hidden temporal correlations and patterns, performing better than conventional methods in both accuracy and interpretability. Moreover, a case study showcasing the application of \"TimeTrail\" in real-world scenarios highlights its utility for fraud detection.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-f1703e0e1610a227", "title": "Integrated Approach of Gearbox Fault Diagnosis", "authors": ["Vikash Kumar", "Subrata Mukherjee", "Somnath Sarangi"], "year": 2023, "published": "2023-08-27T18:35:46Z", "updated": "2023-08-27T18:35:46Z", "arxiv_id": "http://arxiv.org/abs/2308.14174v1", "url": "https://arxiv.org/abs/2308.14174v1", "primary_category": "eess.SP", "categories": ["eess.SP", "cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["phic"], "key_results": ["Gearbox fault diagnosis is one of the most important parts in any industrial systems."], "abstract": "Gearbox fault diagnosis is one of the most important parts in any industrial systems. Failure of components inside gearbox can lead to a catastrophic failure, uneven breakdown, and financial losses in industrial organization. In that case intelligent maintenance of the gearbox comes into context. This paper presents an integrated gearbox fault diagnosis approach which can easily deploy in online condition monitoring. This work introduces a nonparametric data preprocessing technique i.e., calculus enhanced energy operator (CEEO) to preserve the characteristics frequencies in the noisy and inferred vibrational signal. A set of time domain and spectral domain features are calculated from the raw and CEEO vibration signal and inputted to the multiclass support vector machine (MCSVM) to diagnose the faults on the system. An effective comparison between raw signal and CEEO signal are presented to show the impact of CEEO in gearbox fault diagnosis. The obtained results of this work look very promising and can be implemented in any type of industrial system due to its nonparametric nature.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-99d5ab09de03c4a4", "title": "SPEED: Streaming Partition and Parallel Acceleration for Temporal Interaction Graph Embedding", "authors": ["Xi Chen", "Yongxiang Liao", "Yun Xiong", "Yao Zhang", "Siwei Zhang", "Jiawei Zhang", "Yiheng Sun"], "year": 2023, "published": "2023-08-27T15:11:44Z", "updated": "2023-09-11T05:30:49Z", "arxiv_id": "http://arxiv.org/abs/2308.14129v2", "url": "https://arxiv.org/abs/2308.14129v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.DC", "cs.SI"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Our method can achieve a good balance in computing resources, computing time, and downstream task performance."], "abstract": "Temporal Interaction Graphs (TIGs) are widely employed to model intricate real-world systems such as financial systems and social networks. To capture the dynamism and interdependencies of nodes, existing TIG embedding models need to process edges sequentially and chronologically. However, this requirement prevents it from being processed in parallel and struggle to accommodate burgeoning data volumes to GPU. Consequently, many large-scale temporal interaction graphs are confined to CPU processing. Furthermore, a generalized GPU scaling and acceleration approach remains unavailable. To facilitate large-scale TIGs' implementation on GPUs for acceleration, we introduce a novel training approach namely Streaming Edge Partitioning and Parallel Acceleration for Temporal Interaction Graph Embedding (SPEED). The SPEED is comprised of a Streaming Edge Partitioning Component (SEP) which addresses space overhead issue by assigning fewer nodes to each GPU, and a Parallel Acceleration Component (PAC) which enables simultaneous training of different sub-graphs, addressing time overhead issue. Our method can achieve a good balance in computing resources, computing time, and downstream task performance. Empirical validation across 7 real-world datasets demonstrates the potential to expedite training speeds by a factor of up to 19.29x. Simultaneously, resource consumption of a single-GPU can be diminished by up to 69%, thus enabling the multiple GPU-based training and acceleration encompassing millions of nodes and billions of edges. Furthermore, our approach also maintains its competitiveness in downstream tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-388df92c7a891937", "title": "Computation-efficient Deep Learning for Computer Vision: A Survey", "authors": ["Yulin Wang", "Yizeng Han", "Chaofei Wang", "Shiji Song", "Qi Tian", "Gao Huang"], "year": 2023, "published": "2023-08-27T03:55:28Z", "updated": "2023-08-27T03:55:28Z", "arxiv_id": "http://arxiv.org/abs/2308.13998v1", "url": "https://arxiv.org/abs/2308.13998v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.LG", "cs.MM"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["However, the challenge remains that state-of-the-art models usually demand significant computational resources, leading to impractical power consumption, latency, or carbon emissions in real-world scenarios.", "This trade-off between effectiveness and efficiency has catalyzed the emergence of a new research focus: computationally efficient deep learning, which strives to achieve satisfactory performance while minimizing the computational cost during inference."], "abstract": "Over the past decade, deep learning models have exhibited considerable advancements, reaching or even exceeding human-level performance in a range of visual perception tasks. This remarkable progress has sparked interest in applying deep networks to real-world applications, such as autonomous vehicles, mobile devices, robotics, and edge computing. However, the challenge remains that state-of-the-art models usually demand significant computational resources, leading to impractical power consumption, latency, or carbon emissions in real-world scenarios. This trade-off between effectiveness and efficiency has catalyzed the emergence of a new research focus: computationally efficient deep learning, which strives to achieve satisfactory performance while minimizing the computational cost during inference. This review offers an extensive analysis of this rapidly evolving field by examining four key areas: 1) the development of static or dynamic light-weighted backbone models for the efficient extraction of discriminative deep representations; 2) the specialized network architectures or algorithms tailored for specific computer vision tasks; 3) the techniques employed for compressing deep learning models; and 4) the strategies for deploying efficient deep networks on hardware platforms. Additionally, we provide a systematic discussion on the critical challenges faced in this domain, such as network architecture design, training schemes, practical efficiency, and more realistic model compression approaches, as well as potential future research directions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:38:51.631786+00:00"}
{"id": "evg-decee577e6074be1", "title": "The Aegis Protocol: A Foundational Security Framework for Autonomous AI Agents", "authors": ["Sai Teja Reddy Adapala", "Yashwanth Reddy Alugubelly"], "year": 2025, "published": "2025-08-22T06:18:57Z", "updated": "2025-08-22T06:18:57Z", "arxiv_id": "http://arxiv.org/abs/2508.19267v1", "url": "https://arxiv.org/abs/2508.19267v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["phic"], "key_results": ["We formalize an adversary model extending Dolev-Yao for agentic threats and validate the protocol against the STRIDE framework.", "Our quantitative evaluation used a discrete-event simulation, calibrated against cryptographic benchmarks, to model 1,000 agents."], "abstract": "The proliferation of autonomous AI agents marks a paradigm shift toward complex, emergent multi-agent systems. This transition introduces systemic security risks, including control-flow hijacking and cascading failures, that traditional cybersecurity paradigms are ill-equipped to address. This paper introduces the Aegis Protocol, a layered security framework designed to provide strong security guarantees for open agentic ecosystems. The protocol integrates three technological pillars: (1) non-spoofable agent identity via W3C Decentralized Identifiers (DIDs); (2) communication integrity via NIST-standardized post-quantum cryptography (PQC); and (3) verifiable, privacy-preserving policy compliance using the Halo2 zero-knowledge proof (ZKP) system. We formalize an adversary model extending Dolev-Yao for agentic threats and validate the protocol against the STRIDE framework. Our quantitative evaluation used a discrete-event simulation, calibrated against cryptographic benchmarks, to model 1,000 agents. The simulation showed a 0 percent success rate across 20,000 attack trials. For policy verification, analysis of the simulation logs reported a median proof-generation latency of 2.79 seconds, establishing a performance baseline for this class of security. While the evaluation is simulation-based and early-stage, it offers a reproducible baseline for future empirical studies and positions Aegis as a foundation for safe, scalable autonomous AI.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-1e72429b8480f3bb", "title": "IR-Agent: Expert-Inspired LLM Agents for Structure Elucidation from Infrared Spectra", "authors": ["Heewoong Noh", "Namkyeong Lee", "Gyoung S. Na", "Kibum Kim", "Chanyoung Park"], "year": 2025, "published": "2025-08-22T06:07:28Z", "updated": "2026-05-19T02:24:35Z", "arxiv_id": "http://arxiv.org/abs/2508.16112v2", "url": "https://arxiv.org/abs/2508.16112v2", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Through extensive experiments, we demonstrate that IR-Agent not only improves baseline performance on experimental IR spectra but also shows strong adaptability to various forms of chemical information."], "abstract": "Spectral analysis provides crucial clues for the elucidation of unknown materials. Among various techniques, infrared spectroscopy (IR) plays an important role in laboratory settings due to its high accessibility and low cost. However, existing approaches often fail to reflect expert analytical processes and lack flexibility in incorporating diverse types of chemical knowledge, which is essential in real-world analytical scenarios. In this paper, we propose IR-Agent, a novel multi-agent framework for molecular structure elucidation from IR spectra. The framework is designed to emulate expert-driven IR analysis procedures and is inherently extensible. Each agent specializes in a specific aspect of IR interpretation, and their complementary roles enable integrated reasoning, thereby improving the overall accuracy of structure elucidation. Through extensive experiments, we demonstrate that IR-Agent not only improves baseline performance on experimental IR spectra but also shows strong adaptability to various forms of chemical information.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-237b47845d3da525", "title": "Generative Artificial Intelligence and Agents in Research and Teaching", "authors": ["Jussi S. Jauhiainen", "Aurora Toppari"], "year": 2025, "published": "2025-08-22T06:00:45Z", "updated": "2025-08-26T10:23:02Z", "arxiv_id": "http://arxiv.org/abs/2508.16701v2", "url": "https://arxiv.org/abs/2508.16701v2", "primary_category": "cs.CY", "categories": ["cs.CY", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Deep Research"], "benchmarks": [], "models": ["phical"], "key_results": ["This study provides a comprehensive analysis of the development, functioning, and application of generative artificial intelligence (GenAI) and large language models (LLMs), with an emphasis on their implications for research and education."], "abstract": "This study provides a comprehensive analysis of the development, functioning, and application of generative artificial intelligence (GenAI) and large language models (LLMs), with an emphasis on their implications for research and education. It traces the conceptual evolution from artificial intelligence (AI) through machine learning (ML) and deep learning (DL) to transformer architectures, which constitute the foundation of contemporary generative systems. Technical aspects, including prompting strategies, word embeddings, and probabilistic sampling methods (temperature, top-k, and top-p), are examined alongside the emergence of autonomous agents. These elements are considered in relation to both the opportunities they create and the limitations and risks they entail.\n  The work critically evaluates the integration of GenAI across the research process, from ideation and literature review to research design, data collection, analysis, interpretation, and dissemination. While particular attention is given to geographical research, the discussion extends to wider academic contexts. A parallel strand addresses the pedagogical applications of GenAI, encompassing course and lesson design, teaching delivery, assessment, and feedback, with geography education serving as a case example.\n  Central to the analysis are the ethical, social, and environmental challenges posed by GenAI. Issues of bias, intellectual property, governance, and accountability are assessed, alongside the ecological footprint of LLMs and emerging technological strategies for mitigation. The concluding section considers near- and long-term futures of GenAI, including scenarios of sustained adoption, regulation, and potential decline. By situating GenAI within both scholarly practice and educational contexts, the study contributes to critical debates on its transformative potential and societal responsibilities.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-f4a534c9ad5c8ef2", "title": "Consensus Is All You Need: Gossip-Based Reasoning Among Large Language Models", "authors": ["Saksham Arora"], "year": 2025, "published": "2025-08-22T05:49:27Z", "updated": "2025-08-22T05:49:27Z", "arxiv_id": "http://arxiv.org/abs/2508.18292v1", "url": "https://arxiv.org/abs/2508.18292v1", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["Large language models have advanced rapidly, but no single model excels in every area -- each has its strengths and weaknesses."], "abstract": "Large language models have advanced rapidly, but no single model excels in every area -- each has its strengths and weaknesses. Instead of relying on one model alone, we take inspiration from gossip protocols in distributed systems, where information is exchanged with peers until they all come to an agreement. In this setup, models exchange answers and gradually work toward a shared solution. Each LLM acts as a node in a peer-to-peer network, sharing responses and thought processes to reach a collective decision. Our results show that this \"gossip-based consensus\" leads to robust, resilient, and accurate multi-agent AI reasoning. It helps overcome the weaknesses of individual models and brings out their collective strengths. This approach is similar to how humans build consensus, making AI seem more collaborative and trustworthy instead of just a black-box program.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-c2aa8d1ce6eada5c", "title": "X-Troll: eXplainable Detection of State-Sponsored Information Operations Agents", "authors": ["Lin Tian", "Xiuzhen Zhang", "Maria Myung-Hee Kim", "Jennifer Biggs", "Marian-Andrei Rizoiu"], "year": 2025, "published": "2025-08-22T00:52:58Z", "updated": "2025-08-27T02:18:21Z", "arxiv_id": "http://arxiv.org/abs/2508.16021v2", "url": "https://arxiv.org/abs/2508.16021v2", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Interpretability"], "benchmarks": [], "models": ["phisticated"], "key_results": ["While Large Language Models (LLMs) achieve strong performance on general natural language processing (NLP) tasks, they struggle with subtle propaganda detection and operate as ``black boxes'', providing no interpretable insights into manipulation strategies."], "abstract": "State-sponsored trolls, malicious actors who deploy sophisticated linguistic manipulation in coordinated information campaigns, posing threats to online discourse integrity. While Large Language Models (LLMs) achieve strong performance on general natural language processing (NLP) tasks, they struggle with subtle propaganda detection and operate as ``black boxes'', providing no interpretable insights into manipulation strategies. This paper introduces X-Troll, a novel framework that bridges this gap by integrating explainable adapter-based LLMs with expert-derived linguistic knowledge to detect state-sponsored trolls and provide human-readable explanations for its decisions. X-Troll incorporates appraisal theory and propaganda analysis through specialized LoRA adapters, using dynamic gating to capture campaign-specific discourse patterns in coordinated information operations. Experiments on real-world data demonstrate that our linguistically-informed approach shows strong performance compared with both general LLM baselines and existing troll detection models in accuracy while providing enhanced transparency through expert-grounded explanations that reveal the specific linguistic strategies used by state-sponsored actors. X-Troll source code is available at: https://github.com/ltian678/xtroll_source/.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-d11f7313b6875535", "title": "Disentangled Multi-modal Learning of Histology and Transcriptomics for Cancer Characterization", "authors": ["Yupei Zhang", "Xiaofei Wang", "Anran Liu", "Lequan Yu", "Chao Li"], "year": 2025, "published": "2025-08-22T15:51:33Z", "updated": "2026-02-28T05:27:56Z", "arxiv_id": "http://arxiv.org/abs/2508.16479v2", "url": "https://arxiv.org/abs/2508.16479v2", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.AI", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Distillation", "VLM"], "benchmarks": [], "models": [], "key_results": ["4) To improve inference efficiency, we propose an informative token aggregation module that suppresses WSI redundancy while preserving subspace semantics.", "Extensive experiments on cancer diagnosis, prognosis, and survival prediction demonstrate our superiority over state-of-the-art methods across multiple settings."], "abstract": "Histopathology remains the gold standard for cancer diagnosis and prognosis. With the advent of transcriptome profiling, multi-modal learning combining transcriptomics with histology offers more comprehensive information. However, existing multi-modal approaches are challenged by intrinsic multi-modal heterogeneity, insufficient multi-scale integration, and reliance on paired data, restricting clinical applicability. To address these challenges, we propose a disentangled multi-modal framework with four contributions: 1) To mitigate multi-modal heterogeneity, we decompose WSIs and transcriptomes into tumor and microenvironment subspaces using a disentangled multi-modal fusion module, and introduce a confidence-guided gradient coordination strategy to balance subspace optimization. 2) To enhance multi-scale integration, we propose an inter-magnification gene-expression consistency strategy that aligns transcriptomic signals across WSI magnifications. 3) To reduce dependency on paired data, we propose a subspace knowledge distillation strategy enabling transcriptome-agnostic inference through a WSI-only student model. 4) To improve inference efficiency, we propose an informative token aggregation module that suppresses WSI redundancy while preserving subspace semantics. Extensive experiments on cancer diagnosis, prognosis, and survival prediction demonstrate our superiority over state-of-the-art methods across multiple settings. Code is available at https://github.com/helenypzhang/Disentangled-Multimodal-Learning.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-279283f6ac1aa563", "title": "NeuroKoop: Neural Koopman Fusion of Structural-Functional Connectomes for Identifying Prenatal Drug Exposure in Adolescents", "authors": ["Badhan Mazumder", "Aline Kotoski", "Vince D. Calhoun", "Dong Hye Ye"], "year": 2025, "published": "2025-08-22T14:25:19Z", "updated": "2026-01-25T03:21:41Z", "arxiv_id": "http://arxiv.org/abs/2508.16414v2", "url": "https://arxiv.org/abs/2508.16414v2", "primary_category": "q-bio.NC", "categories": ["q-bio.NC", "cs.CV", "eess.IV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Applied to a large adolescent cohort from the ABCD dataset, NeuroKoop outperformed relevant baselines and revealed salient structural-functional connections, advancing our understanding of the neurodevelopmental impact of PDE."], "abstract": "Understanding how prenatal exposure to psychoactive substances such as cannabis shapes adolescent brain organization remains a critical challenge, complicated by the complexity of multimodal neuroimaging data and the limitations of conventional analytic methods. Existing approaches often fail to fully capture the complementary features embedded within structural and functional connectomes, constraining both biological insight and predictive performance. To address this, we introduced NeuroKoop, a novel graph neural network-based framework that integrates structural and functional brain networks utilizing neural Koopman operator-driven latent space fusion. By leveraging Koopman theory, NeuroKoop unifies node embeddings derived from source-based morphometry (SBM) and functional network connectivity (FNC) based brain graphs, resulting in enhanced representation learning and more robust classification of prenatal drug exposure (PDE) status. Applied to a large adolescent cohort from the ABCD dataset, NeuroKoop outperformed relevant baselines and revealed salient structural-functional connections, advancing our understanding of the neurodevelopmental impact of PDE.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-133daf474ca4f54c", "title": "DSRAG: A Domain-Specific Retrieval Framework Based on Document-derived Multimodal Knowledge Graph", "authors": ["Mengzheng Yang", "Yanfei Ren", "David Osei Opoku", "Ruochang Li", "Peng Ren", "Chunxiao Xing"], "year": 2025, "published": "2025-08-22T14:24:48Z", "updated": "2025-08-22T14:24:48Z", "arxiv_id": "http://arxiv.org/abs/2509.10467v1", "url": "https://arxiv.org/abs/2509.10467v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.AI", "cs.CL", "cs.CV", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["Memory / RAG", "VLM"], "benchmarks": [], "models": [], "key_results": ["Current general-purpose large language models (LLMs) commonly exhibit knowledge hallucination and insufficient domain-specific adaptability in domain-specific tasks, limiting their effectiveness in specialized question answering scenarios."], "abstract": "Current general-purpose large language models (LLMs) commonly exhibit knowledge hallucination and insufficient domain-specific adaptability in domain-specific tasks, limiting their effectiveness in specialized question answering scenarios. Retrieval-augmented generation (RAG) effectively tackles these challenges by integrating external knowledge to enhance accuracy and relevance. However, traditional RAG still faces limitations in domain knowledge accuracy and context modeling.To enhance domain-specific question answering performance, this work focuses on a graph-based RAG framework, emphasizing the critical role of knowledge graph quality during the generation process. We propose DSRAG (Domain-Specific RAG), a multimodal knowledge graph-driven retrieval-augmented generation framework designed for domain-specific applications. Our approach leverages domain-specific documents as the primary knowledge source, integrating heterogeneous information such as text, images, and tables to construct a multimodal knowledge graph covering both conceptual and instance layers. Building on this foundation, we introduce semantic pruning and structured subgraph retrieval mechanisms, combining knowledge graph context and vector retrieval results to guide the language model towards producing more reliable responses. Evaluations using the Langfuse multidimensional scoring mechanism show that our method excels in domain-specific question answering, validating the efficacy of integrating multimodal knowledge graphs with retrieval-augmented generation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-882cc3eac198931f", "title": "SAMFusion: Sensor-Adaptive Multimodal Fusion for 3D Object Detection in Adverse Weather", "authors": ["Edoardo Palladin", "Roland Dietze", "Praveen Narayanan", "Mario Bijelic", "Felix Heide"], "year": 2025, "published": "2025-08-22T14:20:46Z", "updated": "2025-08-22T14:20:46Z", "arxiv_id": "http://arxiv.org/abs/2508.16408v1", "url": "https://arxiv.org/abs/2508.16408v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Deep Research", "VLM"], "benchmarks": [], "models": [], "key_results": ["We demonstrate that our method improves the reliability of multimodal sensor fusion in autonomous vehicles under challenging weather conditions, bridging the gap between ideal conditions and real-world edge cases.", "Our approach improves average precision by 17.2 AP compared to the next best method for vulnerable pedestrians in long distances and challenging foggy scenes."], "abstract": "Multimodal sensor fusion is an essential capability for autonomous robots, enabling object detection and decision-making in the presence of failing or uncertain inputs. While recent fusion methods excel in normal environmental conditions, these approaches fail in adverse weather, e.g., heavy fog, snow, or obstructions due to soiling. We introduce a novel multi-sensor fusion approach tailored to adverse weather conditions. In addition to fusing RGB and LiDAR sensors, which are employed in recent autonomous driving literature, our sensor fusion stack is also capable of learning from NIR gated camera and radar modalities to tackle low light and inclement weather. We fuse multimodal sensor data through attentive, depth-based blending schemes, with learned refinement on the Bird's Eye View (BEV) plane to combine image and range features effectively. Our detections are predicted by a transformer decoder that weighs modalities based on distance and visibility. We demonstrate that our method improves the reliability of multimodal sensor fusion in autonomous vehicles under challenging weather conditions, bridging the gap between ideal conditions and real-world edge cases. Our approach improves average precision by 17.2 AP compared to the next best method for vulnerable pedestrians in long distances and challenging foggy scenes. Our project page is available at https://light.princeton.edu/samfusion/", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-35096b505e3d0fc6", "title": "Skills Regularized Task Decomposition for Multi-task Offline Reinforcement Learning", "authors": ["Minjong Yoo", "Sangwoo Cho", "Honguk Woo"], "year": 2024, "published": "2024-08-28T07:36:20Z", "updated": "2024-08-28T07:36:20Z", "arxiv_id": "http://arxiv.org/abs/2408.15593v1", "url": "https://arxiv.org/abs/2408.15593v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["In offline RL where only offline data is used and online interaction with the environment is restricted, it is yet difficult to achieve the optimal policy for multiple tasks, especially when the data quality varies for the tasks.", "To learn the shareable knowledge across those datasets effectively, we employ a task decomposition method for which common skills are jointly learned and used as guidance to reformulate a task in shared and achievable subtasks.", "To improve the performance of offline RL agents learned on the latent space, we also augment datasets with imaginary trajectories relevant to high-quality skills for each task."], "abstract": "Reinforcement learning (RL) with diverse offline datasets can have the advantage of leveraging the relation of multiple tasks and the common skills learned across those tasks, hence allowing us to deal with real-world complex problems efficiently in a data-driven way. In offline RL where only offline data is used and online interaction with the environment is restricted, it is yet difficult to achieve the optimal policy for multiple tasks, especially when the data quality varies for the tasks. In this paper, we present a skill-based multi-task RL technique on heterogeneous datasets that are generated by behavior policies of different quality. To learn the shareable knowledge across those datasets effectively, we employ a task decomposition method for which common skills are jointly learned and used as guidance to reformulate a task in shared and achievable subtasks. In this joint learning, we use Wasserstein auto-encoder (WAE) to represent both skills and tasks on the same latent space and use the quality-weighted loss as a regularization term to induce tasks to be decomposed into subtasks that are more consistent with high-quality skills than others. To improve the performance of offline RL agents learned on the latent space, we also augment datasets with imaginary trajectories relevant to high-quality skills for each task. Through experiments, we show that our multi-task offline RL approach is robust to the mixed configurations of different-quality datasets and it outperforms other state-of-the-art algorithms for several robotic manipulation tasks and drone navigation tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 32183, "fulltext_methods": ["RLVR / GRPO", "World Model"], "fulltext_benchmarks": [], "fulltext_models": ["yield", "ying", "yields", "yielding", "kling sequential decision problems. In conventional RL formulation"], "matched_methods": ["RLVR / GRPO"], "checked_on": "2026-08-18T07:51:18.574858+00:00"}}
{"id": "evg-e04110c6e72bd0da", "title": "SIaM: Self-Improving Code-Assisted Mathematical Reasoning of Large Language Models", "authors": ["Dian Yu", "Baolin Peng", "Ye Tian", "Linfeng Song", "Haitao Mi", "Dong Yu"], "year": 2024, "published": "2024-08-28T06:33:03Z", "updated": "2024-08-28T06:33:03Z", "arxiv_id": "http://arxiv.org/abs/2408.15565v1", "url": "https://arxiv.org/abs/2408.15565v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization"], "benchmarks": ["GSM8K", "MATH"], "models": [], "key_results": ["We also explore different alignment algorithms with self-generated instruction/preference data to foster continuous improvement."], "abstract": "There is a growing trend of teaching large language models (LLMs) to solve mathematical problems through coding. Existing studies primarily focus on prompting powerful, closed-source models to generate seed training data followed by in-domain data augmentation, equipping LLMs with considerable capabilities for code-aided mathematical reasoning. However, continually training these models on augmented data derived from a few datasets such as GSM8K may impair their generalization abilities and restrict their effectiveness to a narrow range of question types. Conversely, the potential of improving such LLMs by leveraging large-scale, expert-written, diverse math question-answer pairs remains unexplored. To utilize these resources and tackle unique challenges such as code response assessment, we propose a novel paradigm that uses a code-based critic model to guide steps including question-code data construction, quality control, and complementary evaluation. We also explore different alignment algorithms with self-generated instruction/preference data to foster continuous improvement. Experiments across both in-domain (up to +5.7%) and out-of-domain (+4.4%) benchmarks in English and Chinese demonstrate the effectiveness of the proposed paradigm.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 63911, "fulltext_methods": ["RLVR / GRPO", "Verifier / PRM", "Chain-of-Thought", "Distillation", "Preference Optimization"], "fulltext_benchmarks": ["GSM8K", "MATH", "DROP"], "fulltext_models": ["GPT-4", "GPT-4-0613", "GPT-4-1106-Preview", "GPT-4-0125-Preview", "gpt-4", "Llama3-8B", "llamaFactory", "Llama-7B-Python", "Llama3", "llama 3"], "matched_methods": ["Preference Optimization"], "checked_on": "2026-08-18T07:51:26.371247+00:00"}}
{"id": "evg-7e45f4f41c1bac9d", "title": "Boosting Lossless Speculative Decoding via Feature Sampling and Partial Alignment Distillation", "authors": ["Lujun Gui", "Bin Xiao", "Lei Su", "Weipeng Chen"], "year": 2024, "published": "2024-08-28T06:28:01Z", "updated": "2024-08-28T06:28:01Z", "arxiv_id": "http://arxiv.org/abs/2408.15562v1", "url": "https://arxiv.org/abs/2408.15562v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Distillation", "Speculative Decoding", "Preference Optimization", "Memory / RAG"], "benchmarks": [], "models": ["LLaMA3-Instruct", "ying"], "key_results": ["In this paper, we reassess these approaches and propose FSPAD (Feature Sampling and Partial Alignment Distillation for Lossless Speculative Decoding), which introduces two straightforward and effective components within the existing framework to boost lossless speculative decoding.", "The results show that FSPAD outperforms the state-of-the-art method across all the aforementioned tasks and target LLMs."], "abstract": "Lossless speculative decoding accelerates target large language model (LLM) inference by employing a lightweight draft model for generating tree-structured candidates, which are subsequently verified in parallel by the target LLM. Currently, effective approaches leverage feature-level rather than token-level autoregression within the draft model to facilitate more straightforward predictions and enhanced knowledge distillation. In this paper, we reassess these approaches and propose FSPAD (Feature Sampling and Partial Alignment Distillation for Lossless Speculative Decoding), which introduces two straightforward and effective components within the existing framework to boost lossless speculative decoding. Firstly, FSPAD utilizes token embeddings to sample features of the target LLM in high-dimensional space before feeding them into the draft model, due to the inherent uncertainty of the features preventing the draft model from obtaining the specific token output by the target LLM. Secondly, FSPAD introduces partial alignment distillation to weaken the draft model's connection between features and logits, aiming to reduce the conflict between feature alignment and logit confidence during training. Our experiments include both greedy and non-greedy decoding on the largest and smallest models from the Vicuna and LLaMA3-Instruct series, as well as tasks in multi-turn conversation, translation, summarization, question answering, mathematical reasoning, and retrieval-augmented generation. The results show that FSPAD outperforms the state-of-the-art method across all the aforementioned tasks and target LLMs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 35231, "fulltext_methods": ["Search / MCTS", "Distillation", "Quantization", "Speculative Decoding", "Preference Optimization", "Memory / RAG"], "fulltext_benchmarks": ["GSM8K", "MT-Bench"], "fulltext_models": ["Gpt-4", "LLaMA3-Instruct", "LLaMA3-Instruct.", "ying"], "matched_methods": ["Distillation", "Memory / RAG", "Preference Optimization", "Speculative Decoding"], "checked_on": "2026-08-18T07:51:31.464244+00:00"}}
{"id": "evg-74f795f46035a719", "title": "Improving Thompson Sampling via Information Relaxation for Budgeted Multi-armed Bandits", "authors": ["Woojin Jeong", "Seungki Min"], "year": 2024, "published": "2024-08-28T04:56:06Z", "updated": "2024-08-28T04:56:06Z", "arxiv_id": "http://arxiv.org/abs/2408.15535v1", "url": "https://arxiv.org/abs/2408.15535v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["In a one-to-one correspondence with these algorithms, a series of performance benchmarks that improve the conventional benchmark are also suggested.", "Our theoretical analysis and simulation results show that our algorithms (and our benchmarks) make incremental improvements over BTS (respectively, the conventional benchmark) across various settings including a real-world example."], "abstract": "We consider a Bayesian budgeted multi-armed bandit problem, in which each arm consumes a different amount of resources when selected and there is a budget constraint on the total amount of resources that can be used. Budgeted Thompson Sampling (BTS) offers a very effective heuristic to this problem, but its arm-selection rule does not take into account the remaining budget information. We adopt \\textit{Information Relaxation Sampling} framework that generalizes Thompson Sampling for classical $K$-armed bandit problems, and propose a series of algorithms that are randomized like BTS but more carefully optimize their decisions with respect to the budget constraint. In a one-to-one correspondence with these algorithms, a series of performance benchmarks that improve the conventional benchmark are also suggested. Our theoretical analysis and simulation results show that our algorithms (and our benchmarks) make incremental improvements over BTS (respectively, the conventional benchmark) across various settings including a real-world example.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 52663, "fulltext_methods": ["RLVR / GRPO", "Deep Research"], "fulltext_benchmarks": [], "fulltext_models": ["ying", "yields", "yield", "Yingce", "ying.", "bert"], "matched_methods": [], "checked_on": "2026-08-18T07:51:40.241759+00:00"}}
{"id": "evg-e8949531cb51c48b", "title": "Continual-learning-based framework for structural damage recognition", "authors": ["Jiangpeng Shu", "Jiawei Zhang", "Reachsak Ly", "Fangzheng Lin", "Yuanfeng Duan"], "year": 2024, "published": "2024-08-28T03:50:04Z", "updated": "2024-08-28T03:50:04Z", "arxiv_id": "http://arxiv.org/abs/2408.15513v1", "url": "https://arxiv.org/abs/2408.15513v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["phic"], "key_results": ["In this way, it reduces both the prediction time and data storage by about 75% in four tasks of continuous learning.", "By gradual feature fusion, CLDRM outperformed other methods by managed to achieve high accuracy in the damage recognition and classification."], "abstract": "Multi-damage is common in reinforced concrete structures and leads to the requirement of large number of neural networks, parameters and data storage, if convolutional neural network (CNN) is used for damage recognition. In addition, conventional CNN experiences catastrophic forgetting and training inefficiency as the number of tasks increases during continual learning, leading to large accuracy decrease of previous learned tasks. To address these problems, this study proposes a continuallearning-based damage recognition model (CLDRM) which integrates the learning without forgetting continual learning method into the ResNet-34 architecture for the recognition of damages in RC structures as well as relevant structural components. Three experiments for four recognition tasks were designed to validate the feasibility and effectiveness of the CLDRM framework. In this way, it reduces both the prediction time and data storage by about 75% in four tasks of continuous learning. Three experiments for four recognition tasks were designed to validate the feasibility and effectiveness of the CLDRM framework. By gradual feature fusion, CLDRM outperformed other methods by managed to achieve high accuracy in the damage recognition and classification. As the number of recognition tasks increased, CLDRM also experienced smaller decrease of the previous learned tasks. Results indicate that the CLDRM framework successfully performs damage recognition and classification with reasonable accuracy and effectiveness.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00", "verified": false, "verification": {"source": "unavailable", "status": "unavailable", "checked_on": "2026-08-18T07:51:56.608891+00:00"}}
{"id": "evg-628c4ef85f0da047", "title": "AeroVerse: UAV-Agent Benchmark Suite for Simulating, Pre-training, Finetuning, and Evaluating Aerospace Embodied World Models", "authors": ["Fanglong Yao", "Yuanchang Yue", "Youzhi Liu", "Xian Sun", "Kun Fu"], "year": 2024, "published": "2024-08-28T03:47:45Z", "updated": "2025-11-20T15:34:13Z", "arxiv_id": "http://arxiv.org/abs/2408.15511v2", "url": "https://arxiv.org/abs/2408.15511v2", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization", "World Model", "VLM"], "benchmarks": [], "models": ["GPT-4", "o10k", "v3k"], "key_results": ["Aerospace embodied intelligence aims to empower unmanned aerial vehicles (UAVs) and other aerospace platforms to achieve autonomous perception, cognition, and action, as well as egocentric active interaction with humans and the environment."], "abstract": "Aerospace embodied intelligence aims to empower unmanned aerial vehicles (UAVs) and other aerospace platforms to achieve autonomous perception, cognition, and action, as well as egocentric active interaction with humans and the environment. The aerospace embodied world model serves as an effective means to realize the autonomous intelligence of UAVs and represents a necessary pathway toward aerospace embodied intelligence. However, existing embodied world models primarily focus on ground-level intelligent agents in indoor scenarios, while research on UAV intelligent agents remains unexplored. To address this gap, we construct the first large-scale real-world image-text pre-training dataset, AerialAgent-Ego10k, featuring urban drones from a first-person perspective. We also create a virtual image-text-pose alignment dataset, CyberAgent Ego500k, to facilitate the pre-training of the aerospace embodied world model. For the first time, we clearly define 5 downstream tasks, i.e., aerospace embodied scene awareness, spatial reasoning, navigational exploration, task planning, and motion decision, and construct corresponding instruction datasets, i.e., SkyAgent-Scene3k, SkyAgent-Reason3k, SkyAgent-Nav3k and SkyAgent-Plan3k, and SkyAgent-Act3k, for fine-tuning the aerospace embodiment world model. Simultaneously, we develop SkyAgentEval, the downstream task evaluation metrics based on GPT-4, to comprehensively, flexibly, and objectively assess the results, revealing the potential and limitations of 2D/3D visual language models in UAV-agent tasks. Furthermore, we integrate over 10 2D/3D visual-language models, 2 pre-training datasets, 5 finetuning datasets, more than 10 evaluation metrics, and a simulator into the benchmark suite, i.e., AeroVerse, which will be released to the community to promote exploration and development of aerospace embodied intelligence.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00", "verified": true, "verification": {"source": "ar5iv", "status": "fulltext-verified", "fulltext_chars": 83148, "fulltext_methods": ["RLVR / GRPO", "Chain-of-Thought", "Preference Optimization", "Interpretability", "World Model", "VLM"], "fulltext_benchmarks": ["MT-Bench", "Chatbot Arena"], "fulltext_models": ["GPT-4", "gpt-4-vision-preview", "gpt-4o", "gpt-4-vision-review", "gpt-4o.", "gpt-4", "GPT-4-vision-preview", "GPT-4o", "Gpt-4", "o10k"], "matched_methods": ["Preference Optimization", "VLM", "World Model"], "checked_on": "2026-08-18T07:52:01.815149+00:00"}}
{"id": "evg-e3856e620d1c8e37", "title": "MODULI: Unlocking Preference Generalization via Diffusion Models for Offline Multi-Objective Reinforcement Learning", "authors": ["Yifu Yuan", "Zhenrui Zheng", "Zibin Dong", "Jianye Hao"], "year": 2024, "published": "2024-08-28T03:10:45Z", "updated": "2025-05-27T03:51:30Z", "arxiv_id": "http://arxiv.org/abs/2408.15501v2", "url": "https://arxiv.org/abs/2408.15501v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["To achieve accurate generation, MODULI introduces two return normalization methods under diverse preferences for refining guidance.", "Extensive experiments on the D4MORL benchmark demonstrate that our algorithm outperforms state-of-the-art Offline MORL baselines, exhibiting excellent generalization to OOD preferences."], "abstract": "Multi-objective Reinforcement Learning (MORL) seeks to develop policies that simultaneously optimize multiple conflicting objectives, but it requires extensive online interactions. Offline MORL provides a promising solution by training on pre-collected datasets to generalize to any preference upon deployment. However, real-world offline datasets are often conservatively and narrowly distributed, failing to comprehensively cover preferences, leading to the emergence of out-of-distribution (OOD) preference areas. Existing offline MORL algorithms exhibit poor generalization to OOD preferences, resulting in policies that do not align with preferences. Leveraging the excellent expressive and generalization capabilities of diffusion models, we propose MODULI (Multi-objective Diffusion Planner with Sliding Guidance), which employs a preference-conditioned diffusion model as a planner to generate trajectories that align with various preferences and derive action for decision-making. To achieve accurate generation, MODULI introduces two return normalization methods under diverse preferences for refining guidance. To further enhance generalization to OOD preferences, MODULI proposes a novel sliding guidance mechanism, which involves training an additional slider adapter to capture the direction of preference changes. Incorporating the slider, it transitions from in-distribution (ID) preferences to generating OOD preferences, patching, and extending the incomplete Pareto front. Extensive experiments on the D4MORL benchmark demonstrate that our algorithm outperforms state-of-the-art Offline MORL baselines, exhibiting excellent generalization to OOD preferences.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-86d93f182490dba2", "title": "Legilimens: Practical and Unified Content Moderation for Large Language Model Services", "authors": ["Jialin Wu", "Jiangyi Deng", "Shengyuan Pang", "Yanjiao Chen", "Jiayang Xu", "Xinfeng Li", "Wenyuan Xu"], "year": 2024, "published": "2024-08-28T02:27:07Z", "updated": "2024-09-05T15:50:44Z", "arxiv_id": "http://arxiv.org/abs/2408.15488v2", "url": "https://arxiv.org/abs/2408.15488v2", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": ["phisticated"], "key_results": ["In this paper, we reveal for the first time that effective and efficient content moderation can be achieved by extracting conceptual features from chat-oriented LLMs, despite their initial fine-tuning for conversation rather than content moderation.", "Our red-team model-based data augmentation enhances the robustness of Legilimens against state-of-the-art jailbreaking.", "We have conducted extensive experiments on five host LLMs, seventeen datasets, and nine jailbreaking methods to verify the effectiveness, efficiency, and robustness of Legilimens against normal and adaptive adversaries."], "abstract": "Given the societal impact of unsafe content generated by large language models (LLMs), ensuring that LLM services comply with safety standards is a crucial concern for LLM service providers. Common content moderation methods are limited by an effectiveness-and-efficiency dilemma, where simple models are fragile while sophisticated models consume excessive computational resources. In this paper, we reveal for the first time that effective and efficient content moderation can be achieved by extracting conceptual features from chat-oriented LLMs, despite their initial fine-tuning for conversation rather than content moderation. We propose a practical and unified content moderation framework for LLM services, named Legilimens, which features both effectiveness and efficiency. Our red-team model-based data augmentation enhances the robustness of Legilimens against state-of-the-art jailbreaking. Additionally, we develop a framework to theoretically analyze the cost-effectiveness of Legilimens compared to other methods. We have conducted extensive experiments on five host LLMs, seventeen datasets, and nine jailbreaking methods to verify the effectiveness, efficiency, and robustness of Legilimens against normal and adaptive adversaries. A comparison of Legilimens with both commercial and academic baselines demonstrates the superior performance of Legilimens. Furthermore, we confirm that Legilimens can be applied to few-shot scenarios and extended to multi-label classification tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-1fcc41646c829201", "title": "Avoiding Generative Model Writer's Block With Embedding Nudging", "authors": ["Ali Zand", "Milad Nasr"], "year": 2024, "published": "2024-08-28T00:07:51Z", "updated": "2024-08-28T00:07:51Z", "arxiv_id": "http://arxiv.org/abs/2408.15450v1", "url": "https://arxiv.org/abs/2408.15450v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Safety / Jailbreak", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Generative image models, since introduction, have become a global phenomenon."], "abstract": "Generative image models, since introduction, have become a global phenomenon. From new arts becoming possible to new vectors of abuse, many new capabilities have become available. One of the challenging issues with generative models is controlling the generation process specially to prevent specific generations classes or instances . There are several reasons why one may want to control the output of generative models, ranging from privacy and safety concerns to application limitations or user preferences\n  To address memorization and privacy challenges, there has been considerable research dedicated to filtering prompts or filtering the outputs of these models. What all these solutions have in common is that at the end of the day they stop the model from producing anything, hence limiting the usability of the model. In this paper, we propose a method for addressing this usability issue by making it possible to steer away from unwanted concepts (when detected in model's output) and still generating outputs. In particular we focus on the latent diffusion image generative models and how one can prevent them to generate particular images while generating similar images with limited overhead.\n  We focus on mitigating issues like image memorization, demonstrating our technique's effectiveness through qualitative and quantitative evaluations. Our method successfully prevents the generation of memorized training images while maintaining comparable image quality and relevance to the unmodified model.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-57b0584a444c1ca2", "title": "Simultaneous Training of First- and Second-Order Optimizers in Population-Based Reinforcement Learning", "authors": ["Felix Pfeiffer", "Shahram Eivazi"], "year": 2024, "published": "2024-08-27T21:54:26Z", "updated": "2024-09-04T10:17:22Z", "arxiv_id": "http://arxiv.org/abs/2408.15421v2", "url": "https://arxiv.org/abs/2408.15421v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["Population-based training (PBT) provides a method to achieve this by continuously tuning hyperparameters throughout the training.", "This ongoing adjustment enables models to adapt to different learning stages, resulting in faster convergence and overall improved performance.", "Specifically, the combination of the K-FAC optimizer with Adam led to up to a 10% improvement in overall performance compared to PBT using only Adam."], "abstract": "The tuning of hyperparameters in reinforcement learning (RL) is critical, as these parameters significantly impact an agent's performance and learning efficiency. Dynamic adjustment of hyperparameters during the training process can significantly enhance both the performance and stability of learning. Population-based training (PBT) provides a method to achieve this by continuously tuning hyperparameters throughout the training. This ongoing adjustment enables models to adapt to different learning stages, resulting in faster convergence and overall improved performance. In this paper, we propose an enhancement to PBT by simultaneously utilizing both first- and second-order optimizers within a single population. We conducted a series of experiments using the TD3 algorithm across various MuJoCo environments. Our results, for the first time, empirically demonstrate the potential of incorporating second-order optimizers within PBT-based RL. Specifically, the combination of the K-FAC optimizer with Adam led to up to a 10% improvement in overall performance compared to PBT using only Adam. Additionally, in environments where Adam occasionally fails, such as the Swimmer environment, the mixed population with K-FAC exhibited more reliable learning outcomes, offering a significant advantage in training stability without a substantial increase in computational time.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-b1891e66c13b7e28", "title": "Awes, Laws, and Flaws From Today's LLM Research", "authors": ["Adrian de Wynter"], "year": 2024, "published": "2024-08-27T21:19:37Z", "updated": "2025-05-30T22:55:40Z", "arxiv_id": "http://arxiv.org/abs/2408.15409v3", "url": "https://arxiv.org/abs/2408.15409v3", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["We perform a critical examination of the scientific methodology behind contemporary large language model (LLM) research."], "abstract": "We perform a critical examination of the scientific methodology behind contemporary large language model (LLM) research. For this we assess over 2,000 research works released between 2020 and 2024 based on criteria typical of what is considered good research (e.g. presence of statistical tests and reproducibility), and cross-validate it with arguments that are at the centre of controversy (e.g., claims of emergent behaviour). We find multiple trends, such as declines in ethics disclaimers, a rise of LLMs as evaluators, and an increase on claims of LLM reasoning abilities without leveraging human evaluation. We note that conference checklists are effective at curtailing some of these issues, but balancing velocity and rigour in research cannot solely rely on these. We tie all these findings to findings from recent meta-reviews and extend recommendations on how to address what does, does not, and should work in LLM research.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-5761882f0300aae2", "title": "On Stateful Value Factorization in Multi-Agent Reinforcement Learning", "authors": ["Enrico Marchesini", "Andrea Baisero", "Rupali Bhati", "Christopher Amato"], "year": 2024, "published": "2024-08-27T19:45:26Z", "updated": "2024-09-09T22:49:17Z", "arxiv_id": "http://arxiv.org/abs/2408.15381v2", "url": "https://arxiv.org/abs/2408.15381v2", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["We then introduce DuelMIX, a factorization algorithm that learns distinct per-agent utility estimators to improve performance and achieve full expressiveness."], "abstract": "Value factorization is a popular paradigm for designing scalable multi-agent reinforcement learning algorithms. However, current factorization methods make choices without full justification that may limit their performance. For example, the theory in prior work uses stateless (i.e., history) functions, while the practical implementations use state information -- making the motivating theory a mismatch for the implementation. Also, methods have built off of previous approaches, inheriting their architectures without exploring other, potentially better ones. To address these concerns, we formally analyze the theory of using the state instead of the history in current methods -- reconnecting theory and practice. We then introduce DuelMIX, a factorization algorithm that learns distinct per-agent utility estimators to improve performance and achieve full expressiveness. Experiments on StarCraft II micromanagement and Box Pushing tasks demonstrate the benefits of our intuitions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-91698e2678a7381f", "title": "Wait, that's not an option: LLMs Robustness with Incorrect Multiple-Choice Options", "authors": ["Gracjan Góral", "Emilia Wiśnios", "Piotr Sankowski", "Paweł Budzianowski"], "year": 2024, "published": "2024-08-27T19:27:43Z", "updated": "2025-06-02T09:08:56Z", "arxiv_id": "http://arxiv.org/abs/2409.00113v3", "url": "https://arxiv.org/abs/2409.00113v3", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Through systematic evaluation across arithmetic, domain-specific knowledge, and high-stakes medical decision tasks, we demonstrate that post-training aligned models often default to selecting invalid options, while base models exhibit improved refusal capabilities that scale with model size."], "abstract": "This work introduces a novel framework for evaluating LLMs' capacity to balance instruction-following with critical reasoning when presented with multiple-choice questions containing no valid answers. Through systematic evaluation across arithmetic, domain-specific knowledge, and high-stakes medical decision tasks, we demonstrate that post-training aligned models often default to selecting invalid options, while base models exhibit improved refusal capabilities that scale with model size. Our analysis reveals that alignment techniques, though intended to enhance helpfulness, can inadvertently impair models' reflective judgment--the ability to override default behaviors when faced with invalid options. We additionally conduct a parallel human study showing similar instruction-following biases, with implications for how these biases may propagate through human feedback datasets used in alignment. We provide extensive ablation studies examining the impact of model size, training techniques, and prompt engineering. Our findings highlight fundamental tensions between alignment optimization and preservation of critical reasoning capabilities, with important implications for developing more robust AI systems for real-world deployment.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-c776f464194772a2", "title": "Optimization Solution Functions as Deterministic Policies for Offline Reinforcement Learning", "authors": ["Vanshaj Khattar", "Ming Jin"], "year": 2024, "published": "2024-08-27T19:04:32Z", "updated": "2024-08-27T19:04:32Z", "arxiv_id": "http://arxiv.org/abs/2408.15368v1", "url": "https://arxiv.org/abs/2408.15368v1", "primary_category": "cs.LG", "categories": ["cs.LG", "eess.SY"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": ["ying"], "key_results": ["Finally, we validate the proposed framework on two real-world applications and show a significant improvement over state-of-the-art (SOTA) offline RL methods."], "abstract": "Offline reinforcement learning (RL) is a promising approach for many control applications but faces challenges such as limited data coverage and value function overestimation. In this paper, we propose an implicit actor-critic (iAC) framework that employs optimization solution functions as a deterministic policy (actor) and a monotone function over the optimal value of optimization as a critic. By encoding optimality in the actor policy, we show that the learned policies are robust to the suboptimality of the learned actor parameters via the exponentially decaying sensitivity (EDS) property. We obtain performance guarantees for the proposed iAC framework and show its benefits over general function approximation schemes. Finally, we validate the proposed framework on two real-world applications and show a significant improvement over state-of-the-art (SOTA) offline RL methods.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-2419acc4c6cb677f", "title": "What makes math problems hard for reinforcement learning: a case study", "authors": ["Ali Shehper", "Anibal M. Medina-Mardones", "Lucas Fagan", "Bartłomiej Lewandowski", "Angus Gruen", "Yang Qiu", "Piotr Kucharski", "Zhenghan Wang", "Sergei Gukov"], "year": 2024, "published": "2024-08-27T18:00:06Z", "updated": "2025-02-11T18:01:40Z", "arxiv_id": "http://arxiv.org/abs/2408.15332v2", "url": "https://arxiv.org/abs/2408.15332v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "math.CO", "math.GR", "math.GT"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["Using a long-standing conjecture from combinatorial group theory, we explore, from multiple perspectives, the challenges of finding rare instances carrying disproportionately high rewards."], "abstract": "Using a long-standing conjecture from combinatorial group theory, we explore, from multiple perspectives, the challenges of finding rare instances carrying disproportionately high rewards. Based on lessons learned in the context defined by the Andrews-Curtis conjecture, we propose algorithmic enhancements and a topological hardness measure with implications for a broad class of search problems. As part of our study, we also address several open mathematical questions. Notably, we demonstrate the length reducibility of all but two presentations in the Akbulut-Kirby series (1981), and resolve various potential counterexamples in the Miller-Schupp series (1991), including three infinite subfamilies.", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-3ab869fd0cb25a70", "title": "Artificially intelligent Maxwell's demon for optimal control of open quantum systems", "authors": ["Paolo Andrea Erdman", "Robert Czupryniak", "Bibek Bhandari", "Andrew N. Jordan", "Frank Noé", "Jens Eisert", "Giacomo Guarnieri"], "year": 2024, "published": "2024-08-27T18:00:02Z", "updated": "2024-08-27T18:00:02Z", "arxiv_id": "http://arxiv.org/abs/2408.15328v1", "url": "https://arxiv.org/abs/2408.15328v1", "primary_category": "quant-ph", "categories": ["quant-ph", "cond-mat.mes-hall", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Interpretability"], "benchmarks": [], "models": [], "key_results": ["Finally, we study the case when all timescales are comparable, finding new feedback control strategies that considerably outperform more intuitive ones."], "abstract": "Feedback control of open quantum systems is of fundamental importance for practical applications in various contexts, ranging from quantum computation to quantum error correction and quantum metrology. Its use in the context of thermodynamics further enables the study of the interplay between information and energy. However, deriving optimal feedback control strategies is highly challenging, as it involves the optimal control of open quantum systems, the stochastic nature of quantum measurement, and the inclusion of policies that maximize a long-term time- and trajectory-averaged goal. In this work, we employ a reinforcement learning approach to automate and capture the role of a quantum Maxwell's demon: the agent takes the literal role of discovering optimal feedback control strategies in qubit-based systems that maximize a trade-off between measurement-powered cooling and measurement efficiency. Considering weak or projective quantum measurements, we explore different regimes based on the ordering between the thermalization, the measurement, and the unitary feedback timescales, finding different and highly non-intuitive, yet interpretable, strategies. In the thermalization-dominated regime, we find strategies with elaborate finite-time thermalization protocols conditioned on measurement outcomes. In the measurement-dominated regime, we find that optimal strategies involve adaptively measuring different qubit observables reflecting the acquired information, and repeating multiple weak measurements until the quantum state is \"sufficiently pure\", leading to random walks in state space. Finally, we study the case when all timescales are comparable, finding new feedback control strategies that considerably outperform more intuitive ones. We discuss a two-qubit example where we explore the role of entanglement and conclude discussing the scaling of our results to quantum many-body systems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4e39401a6f8a195d", "title": "Generative Verifiers: Reward Modeling as Next-Token Prediction", "authors": ["Lunjun Zhang", "Arian Hosseini", "Hritik Bansal", "Mehran Kazemi", "Aviral Kumar", "Rishabh Agarwal"], "year": 2024, "published": "2024-08-27T17:57:45Z", "updated": "2025-02-22T10:21:46Z", "arxiv_id": "http://arxiv.org/abs/2408.15240v3", "url": "https://arxiv.org/abs/2408.15240v3", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Search / MCTS", "Verifier / PRM", "Chain-of-Thought", "Test-time Scaling", "Preference Optimization"], "benchmarks": ["MMLU", "GSM8K", "MATH"], "models": [], "key_results": ["We demonstrate that GenRM outperforms discriminative, DPO verifiers, and LLM-as-a-Judge, resulting in large performance gains with Best-of-N, namely 5% $\\rightarrow$ 45.3% on algorithmic tasks and 73% $\\rightarrow$ 93.4% on GSM8K.", "In easy-to-hard generalization settings, we observe improvements of 28% $\\rightarrow$ 44.6% on MATH, and 37.9% $\\rightarrow$ 53.5% on MMLU abstract algebra."], "abstract": "Verifiers or reward models are often used to enhance the reasoning performance of large language models (LLMs). A common approach is the Best-of-N method, where N candidate solutions generated by the LLM are ranked by a verifier, and the best one is selected. While LLM-based verifiers are typically trained as discriminative classifiers to score solutions, they do not utilize the text generation capabilities of pretrained LLMs. To overcome this limitation, we instead propose training verifiers using the ubiquitous next-token prediction objective, jointly on verification and solution generation. Compared to standard verifiers, such generative verifiers (GenRM) can benefit from several advantages of LLMs: they integrate seamlessly with instruction tuning, enable chain-of-thought reasoning, and can utilize additional test-time compute via majority voting for better verification. We demonstrate that GenRM outperforms discriminative, DPO verifiers, and LLM-as-a-Judge, resulting in large performance gains with Best-of-N, namely 5% $\\rightarrow$ 45.3% on algorithmic tasks and 73% $\\rightarrow$ 93.4% on GSM8K. In easy-to-hard generalization settings, we observe improvements of 28% $\\rightarrow$ 44.6% on MATH, and 37.9% $\\rightarrow$ 53.5% on MMLU abstract algebra. Furthermore, we find that training GenRM with synthetic verification rationales is sufficient to pick out subtle errors on math problems. Finally, we demonstrate that GenRM scales favorably with model size and test-time compute.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-ca73a902c61f9c90", "title": "Bi-Factorial Preference Optimization: Balancing Safety-Helpfulness in Language Models", "authors": ["Wenxuan Zhang", "Philip H. S. Torr", "Mohamed Elhoseiny", "Adel Bibi"], "year": 2024, "published": "2024-08-27T17:31:21Z", "updated": "2025-04-08T11:04:33Z", "arxiv_id": "http://arxiv.org/abs/2408.15313v2", "url": "https://arxiv.org/abs/2408.15313v2", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["The results indicate that our method significantly outperforms existing approaches in both safety and helpfulness.", "Moreover, BFPO achieves the same level of safety as methods that heavily rely on human labor with less than 10\\% of the computational resources and human prompting and annotation process."], "abstract": "Fine-tuning large language models (LLMs) on human preferences, typically through reinforcement learning from human feedback (RLHF), has proven successful in enhancing their capabilities. However, ensuring the safety of LLMs during fine-tuning remains a critical concern, and mitigating the potential conflicts in safety and helpfulness is costly in RLHF. To address this issue, we propose a supervised learning framework called Bi-Factorial Preference Optimization (BFPO), which re-parameterizes a joint RLHF objective of both safety and helpfulness into a single supervised learning objective. In supervised optimization, a labeling function is used to capture the global preferences ranking to balance both safety and helpfulness. To evaluate BFPO, we develop a benchmark that includes comprehensive discriminative and generative tasks for helpfulness and harmlessness. The results indicate that our method significantly outperforms existing approaches in both safety and helpfulness. Moreover, BFPO achieves the same level of safety as methods that heavily rely on human labor with less than 10\\% of the computational resources and human prompting and annotation process. The training recipes can be found here: https://github.com/wx-zhang/bfpo.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-bfe2078d56294168", "title": "Unifying Symbolic Music Arrangement: Track-Aware Reconstruction and Structured Tokenization", "authors": ["Longshen Ou", "Jingwei Zhao", "Ziyu Wang", "Gus Xia", "Qihao Liang", "Torin Hopkins Ye Wang"], "year": 2024, "published": "2024-08-27T16:18:51Z", "updated": "2025-11-05T08:24:17Z", "arxiv_id": "http://arxiv.org/abs/2408.15176v5", "url": "https://arxiv.org/abs/2408.15176v5", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.CL", "eess.AS"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Our method outperforms task-specific state-of-the-art models on representative tasks in different arrangement scenarios -- band arrangement, piano reduction, and drum arrangement, in both objective metrics and perceptual evaluations."], "abstract": "We present a unified framework for automatic multitrack music arrangement that enables a single pre-trained symbolic music model to handle diverse arrangement scenarios, including reinterpretation, simplification, and additive generation. At its core is a segment-level reconstruction objective operating on token-level disentangled content and style, allowing for flexible any-to-any instrumentation transformations at inference time. To support track-wise modeling, we introduce REMI-z, a structured tokenization scheme for multitrack symbolic music that enhances modeling efficiency and effectiveness for both arrangement tasks and unconditional generation. Our method outperforms task-specific state-of-the-art models on representative tasks in different arrangement scenarios -- band arrangement, piano reduction, and drum arrangement, in both objective metrics and perceptual evaluations. Taken together, our framework demonstrates strong generality and suggests broader applicability in symbolic music-to-music transformation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-ed5913e65f0220fa", "title": "Exploiting Approximate Symmetry for Efficient Multi-Agent Reinforcement Learning", "authors": ["Batuhan Yardim", "Niao He"], "year": 2024, "published": "2024-08-27T16:11:20Z", "updated": "2024-08-27T16:11:20Z", "arxiv_id": "http://arxiv.org/abs/2408.15173v1", "url": "https://arxiv.org/abs/2408.15173v1", "primary_category": "cs.GT", "categories": ["cs.GT", "cs.LG", "math.OC", "stat.ML"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Multi-Agent"], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["Mean-field games (MFG) have become significant tools for solving large-scale multi-agent reinforcement learning problems under symmetry."], "abstract": "Mean-field games (MFG) have become significant tools for solving large-scale multi-agent reinforcement learning problems under symmetry. However, the assumption of exact symmetry limits the applicability of MFGs, as real-world scenarios often feature inherent heterogeneity. Furthermore, most works on MFG assume access to a known MFG model, which might not be readily available for real-world finite-agent games. In this work, we broaden the applicability of MFGs by providing a methodology to extend any finite-player, possibly asymmetric, game to an \"induced MFG\". First, we prove that $N$-player dynamic games can be symmetrized and smoothly extended to the infinite-player continuum via explicit Kirszbraun extensions. Next, we propose the notion of $α,β$-symmetric games, a new class of dynamic population games that incorporate approximate permutation invariance. For $α,β$-symmetric games, we establish explicit approximation bounds, demonstrating that a Nash policy of the induced MFG is an approximate Nash of the $N$-player dynamic game. We show that TD learning converges up to a small bias using trajectories of the $N$-player game with finite-sample guarantees, permitting symmetrized learning without building an explicit MFG model. Finally, for certain games satisfying monotonicity, we prove a sample complexity of $\\widetilde{\\mathcal{O}}(\\varepsilon^{-6})$ for the $N$-agent game to learn an $\\varepsilon$-Nash up to symmetrization bias. Our theory is supported by evaluations on MARL benchmarks with thousands of agents.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-f99e959e4384d1e6", "title": "Using LLMs for Explaining Sets of Counterfactual Examples to Final Users", "authors": ["Arturo Fredes", "Jordi Vitria"], "year": 2024, "published": "2024-08-27T15:13:06Z", "updated": "2024-08-27T15:13:06Z", "arxiv_id": "http://arxiv.org/abs/2408.15133v1", "url": "https://arxiv.org/abs/2408.15133v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Causality is vital for understanding true cause-and-effect relationships between variables within predictive models, rather than relying on mere correlations, making it highly relevant in the field of Explainable AI."], "abstract": "Causality is vital for understanding true cause-and-effect relationships between variables within predictive models, rather than relying on mere correlations, making it highly relevant in the field of Explainable AI. In an automated decision-making scenario, causal inference methods can analyze the underlying data-generation process, enabling explanations of a model's decision by manipulating features and creating counterfactual examples. These counterfactuals explore hypothetical scenarios where a minimal number of factors are altered, providing end-users with valuable information on how to change their situation. However, interpreting a set of multiple counterfactuals can be challenging for end-users who are not used to analyzing raw data records. In our work, we propose a novel multi-step pipeline that uses counterfactuals to generate natural language explanations of actions that will lead to a change in outcome in classifiers of tabular data using LLMs. This pipeline is designed to guide the LLM through smaller tasks that mimic human reasoning when explaining a decision based on counterfactual cases. We conducted various experiments using a public dataset and proposed a method of closed-loop evaluation to assess the coherence of the final explanation with the counterfactuals, as well as the quality of the content. Results are promising, although further experiments with other datasets and human evaluations should be carried out.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-8fa36116b8d556c7", "title": "Evaluating the Energy Consumption of Machine Learning: Systematic Literature Review and Experiments", "authors": ["Charlotte Rodriguez", "Laura Degioanni", "Laetitia Kameni", "Richard Vidal", "Giovanni Neglia"], "year": 2024, "published": "2024-08-27T15:08:06Z", "updated": "2024-08-27T15:08:06Z", "arxiv_id": "http://arxiv.org/abs/2408.15128v1", "url": "https://arxiv.org/abs/2408.15128v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CY"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Deep Research"], "benchmarks": [], "models": [], "key_results": ["Monitoring, understanding, and optimizing the energy consumption of Machine Learning (ML) are various reasons why it is necessary to evaluate the energy usage of ML."], "abstract": "Monitoring, understanding, and optimizing the energy consumption of Machine Learning (ML) are various reasons why it is necessary to evaluate the energy usage of ML. However, there exists no universal tool that can answer this question for all use cases, and there may even be disagreement on how to evaluate energy consumption for a specific use case. Tools and methods are based on different approaches, each with their own advantages and drawbacks, and they need to be mapped out and explained in order to select the most suitable one for a given situation. We address this challenge through two approaches. First, we conduct a systematic literature review of all tools and methods that permit to evaluate the energy consumption of ML (both at training and at inference), irrespective of whether they were originally designed for machine learning or general software. Second, we develop and use an experimental protocol to compare a selection of these tools and methods. The comparison is both qualitative and quantitative on a range of ML tasks of different nature (vision, language) and computational complexity. The systematic literature review serves as a comprehensive guide for understanding the array of tools and methods used in evaluating energy consumption of ML, for various use cases going from basic energy monitoring to consumption optimization. Two open-source repositories are provided for further exploration. The first one contains tools that can be used to replicate this work or extend the current review. The second repository houses the experimental protocol, allowing users to augment the protocol with new ML computing tasks and additional energy evaluation tools.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-b0657347026a6059", "title": "Force-Guided Bridge Matching for Full-Atom Time-Coarsened Dynamics of Peptides", "authors": ["Ziyang Yu", "Wenbing Huang", "Yang Liu"], "year": 2024, "published": "2024-08-27T15:07:27Z", "updated": "2024-12-04T03:24:18Z", "arxiv_id": "http://arxiv.org/abs/2408.15126v6", "url": "https://arxiv.org/abs/2408.15126v6", "primary_category": "physics.chem-ph", "categories": ["physics.chem-ph", "cs.LG", "physics.comp-ph", "q-bio.BM"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Molecular Dynamics (MD) is crucial in various fields such as materials science, chemistry, and pharmacology to name a few."], "abstract": "Molecular Dynamics (MD) is crucial in various fields such as materials science, chemistry, and pharmacology to name a few. Conventional MD software struggles with the balance between time cost and prediction accuracy, which restricts its wider application. Recently, data-driven approaches based on deep generative models have been devised for time-coarsened dynamics, which aim at learning dynamics of diverse molecular systems over a long timestep, enjoying both universality and efficiency. Nevertheless, most current methods are designed solely to learn from the data distribution regardless of the underlying Boltzmann distribution, and the physics priors such as energies and forces are constantly overlooked. In this work, we propose a conditional generative model called Force-guided Bridge Matching (FBM), which learns full-atom time-coarsened dynamics and targets the Boltzmann-constrained distribution. With the guidance of our delicately-designed intermediate force field, FBM leverages favourable physics priors into the generation process, giving rise to enhanced simulations. Experiments on two datasets consisting of peptides verify our superiority in terms of comprehensive metrics and demonstrate transferability to unseen systems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-e9de26e61155066b", "title": "TrafficGamer: Reliable and Flexible Traffic Simulation for Safety-Critical Scenarios with Game-Theoretic Oracles", "authors": ["Guanren Qiao", "Guorui Quan", "Jiawei Yu", "Shujun Jia", "Guiliang Liu"], "year": 2024, "published": "2024-08-28T05:11:16Z", "updated": "2025-12-17T06:52:21Z", "arxiv_id": "http://arxiv.org/abs/2408.15538v3", "url": "https://arxiv.org/abs/2408.15538v3", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Safety / Jailbreak", "Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["While modern Autonomous Vehicle (AV) systems can develop reliable driving policies under regular traffic conditions, they frequently struggle with safety-critical traffic scenarios."], "abstract": "While modern Autonomous Vehicle (AV) systems can develop reliable driving policies under regular traffic conditions, they frequently struggle with safety-critical traffic scenarios. This difficulty primarily arises from the rarity of such scenarios in driving datasets and the complexities associated with predictive modeling of multiple vehicles. Effectively simulating safety-critical traffic situations is therefore a crucial challenge. In this paper, we introduce TrafficGamer, which facilitates game-theoretic traffic simulation by viewing common road driving as a multi-agent game. When we evaluate the empirical performance across various real-world datasets, TrafficGamer ensures both the fidelity, exploitability, and diversity of the simulated scenarios, guaranteeing that they not only statically align with real-world traffic distribution but also efficiently capture equilibria for representing safety-critical scenarios involving multiple agents compared with other methods. Additionally, the results demonstrate that TrafficGamer provides highly flexible simulations across various contexts. Specifically, we demonstrate that the generated scenarios can dynamically adapt to equilibria of varying tightness by configuring risk-sensitive constraints during optimization. We have provided a demo webpage at: https://anonymous.4open.science/api/repo/trafficgamer-demo-1EE0/file/index.html.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-63ac5a22a1acd466", "title": "Toward Automated Simulation Research Workflow through LLM Prompt Engineering Design", "authors": ["Zhihan Liu", "Yubo Chai", "Jianfeng Li"], "year": 2024, "published": "2024-08-28T03:48:05Z", "updated": "2025-01-15T09:12:02Z", "arxiv_id": "http://arxiv.org/abs/2408.15512v3", "url": "https://arxiv.org/abs/2408.15512v3", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "physics.chem-ph"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["GPT-4o", "Claude-3.5"], "key_results": ["Our findings revealed that ASA-GPT-4o achieved near-flawless execution on designated research missions, underscoring the potential of methods like ASA to achieve automation in simulation research processes to enhance research efficiency."], "abstract": "The advent of Large Language Models (LLMs) has created new opportunities for the automation of scientific research spanning both experimental processes and computational simulations. This study explores the feasibility of constructing an autonomous simulation agent (ASA) powered by LLMs through prompt engineering and automated program design to automate the entire simulation research process according to a human-provided research plan. This process includes experimental design, remote upload and simulation execution, data analysis, and report compilation. Using a well-studied simulation problem of polymer chain conformations as a test case, we assessed the long-task completion and reliability of ASAs powered by different LLMs, including GPT-4o, Claude-3.5, etc. Our findings revealed that ASA-GPT-4o achieved near-flawless execution on designated research missions, underscoring the potential of methods like ASA to achieve automation in simulation research processes to enhance research efficiency. The outlined automation can be iteratively performed for up to 20 cycles without human intervention, illustrating the potential of ASA for long-task workflow automation. Additionally, we discussed the intrinsic traits of ASA in managing extensive tasks, focusing on self-validation mechanisms, and the balance between local attention and global oversight.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-edb70393c13f5c35", "title": "RoboSense: Large-scale Dataset and Benchmark for Egocentric Robot Perception and Navigation in Crowded and Unstructured Environments", "authors": ["Haisheng Su", "Feixiang Song", "Cong Ma", "Wei Wu", "Junchi Yan"], "year": 2024, "published": "2024-08-28T03:17:40Z", "updated": "2025-03-05T05:14:34Z", "arxiv_id": "http://arxiv.org/abs/2408.15503v5", "url": "https://arxiv.org/abs/2408.15503v5", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Reliable embodied perception from an egocentric perspective is challenging yet essential for autonomous navigation technology of intelligent mobile agents."], "abstract": "Reliable embodied perception from an egocentric perspective is challenging yet essential for autonomous navigation technology of intelligent mobile agents. With the growing demand of social robotics, near-field scene understanding becomes an important research topic in the areas of egocentric perceptual tasks related to navigation in both crowded and unstructured environments. Due to the complexity of environmental conditions and difficulty of surrounding obstacles owing to truncation and occlusion, the perception capability under this circumstance is still inferior. To further enhance the intelligence of mobile robots, in this paper, we setup an egocentric multi-sensor data collection platform based on 3 main types of sensors (Camera, LiDAR and Fisheye), which supports flexible sensor configurations to enable dynamic sight of view from ego-perspective, capturing either near or farther areas. Meanwhile, a large-scale multimodal dataset is constructed, named RoboSense, to facilitate egocentric robot perception. Specifically, RoboSense contains more than 133K synchronized data with 1.4M 3D bounding box and IDs annotated in the full $360^{\\circ}$ view, forming 216K trajectories across 7.6K temporal sequences. It has $270\\times$ and $18\\times$ as many annotations of surrounding obstacles within near ranges as the previous datasets collected for autonomous driving scenarios such as KITTI and nuScenes. Moreover, we define a novel matching criterion for near-field 3D perception and prediction metrics. Based on RoboSense, we formulate 6 popular tasks to facilitate the future research development, where the detailed analysis as well as benchmarks are also provided accordingly. Data desensitization measures have been conducted for privacy protection.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-346d23ad947ca79a", "title": "Graph Attention Inference of Network Topology in Multi-Agent Systems", "authors": ["Akshay Kolli", "Reza Azadeh", "Kshitj Jerath"], "year": 2024, "published": "2024-08-27T23:58:51Z", "updated": "2024-10-26T19:14:33Z", "arxiv_id": "http://arxiv.org/abs/2408.15449v2", "url": "https://arxiv.org/abs/2408.15449v2", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["Our results demonstrate that the presented data-driven graph attention machine learning model can identify the network topology in multi-agent systems, even when the underlying dynamic model is not known, as evidenced by the F1 scores achieved in the link prediction."], "abstract": "Accurately identifying the underlying graph structures of multi-agent systems remains a difficult challenge. Our work introduces a novel machine learning-based solution that leverages the attention mechanism to predict future states of multi-agent systems by learning node representations. The graph structure is then inferred from the strength of the attention values. This approach is applied to both linear consensus dynamics and the non-linear dynamics of Kuramoto oscillators, resulting in implicit learning of the graph by learning good agent representations. Our results demonstrate that the presented data-driven graph attention machine learning model can identify the network topology in multi-agent systems, even when the underlying dynamic model is not known, as evidenced by the F1 scores achieved in the link prediction.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-fecddaed8763a3f0", "title": "Fast and Modular Autonomy Software for Autonomous Racing Vehicles", "authors": ["Andrew Saba", "Aderotimi Adetunji", "Adam Johnson", "Aadi Kothari", "Matthew Sivaprakasam", "Joshua Spisak", "Prem Bharatia", "Arjun Chauhan", "Brendan Duff", "Noah Gasparro", "Charles King", "Ryan Larkin", "Brian Mao", "Micah Nye", "Anjali Parashar", "Joseph Attias", "Aurimas Balciunas", "Austin Brown", "Chris Chang", "Ming Gao", "Cindy Heredia", "Andrew Keats", "Jose Lavariega", "William Muckelroy", "Andre Slavescu", "Nickolas Stathas", "Nayana Suvarna", "Chuan Tian Zhang", "Sebastian Scherer", "Deva Ramanan"], "year": 2024, "published": "2024-08-27T21:57:16Z", "updated": "2024-08-27T21:57:16Z", "arxiv_id": "http://arxiv.org/abs/2408.15425v1", "url": "https://arxiv.org/abs/2408.15425v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.SE"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["We also cover what did and did not work when deployed on a physical system the Dallara AV-21 platform and potential improvements to address these shortcomings.", "Finally, we convey lessons learned and discuss limitations and future directions for improvement."], "abstract": "Autonomous motorsports aim to replicate the human racecar driver with software and sensors. As in traditional motorsports, Autonomous Racing Vehicles (ARVs) are pushed to their handling limits in multi-agent scenarios at extremely high ($\\geq 150mph$) speeds. This Operational Design Domain (ODD) presents unique challenges across the autonomy stack. The Indy Autonomous Challenge (IAC) is an international competition aiming to advance autonomous vehicle development through ARV competitions. While far from challenging what a human racecar driver can do, the IAC is pushing the state of the art by facilitating full-sized ARV competitions. This paper details the MIT-Pitt-RW Team's approach to autonomous racing in the IAC. In this work, we present our modular and fast approach to agent detection, motion planning and controls to create an autonomy stack. We also provide analysis of the performance of the software stack in single and multi-agent scenarios for rapid deployment in a fast-paced competition environment. We also cover what did and did not work when deployed on a physical system the Dallara AV-21 platform and potential improvements to address these shortcomings. Finally, we convey lessons learned and discuss limitations and future directions for improvement.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7dec830687cec0a9", "title": "Reflective Human-Machine Co-adaptation for Enhanced Text-to-Image Generation Dialogue System", "authors": ["Yuheng Feng", "Yangfan He", "Yinghui Xia", "Tianyu Shi", "Jun Wang", "Jinsong Yang"], "year": 2024, "published": "2024-08-27T18:08:00Z", "updated": "2024-08-27T18:08:00Z", "arxiv_id": "http://arxiv.org/abs/2409.07464v1", "url": "https://arxiv.org/abs/2409.07464v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["To achieve this, we propose a reflective human-machine co-adaptation strategy, named RHM-CAS."], "abstract": "Today's image generation systems are capable of producing realistic and high-quality images. However, user prompts often contain ambiguities, making it difficult for these systems to interpret users' potential intentions. Consequently, machines need to interact with users multiple rounds to better understand users' intents. The unpredictable costs of using or learning image generation models through multiple feedback interactions hinder their widespread adoption and full performance potential, especially for non-expert users. In this research, we aim to enhance the user-friendliness of our image generation system. To achieve this, we propose a reflective human-machine co-adaptation strategy, named RHM-CAS. Externally, the Agent engages in meaningful language interactions with users to reflect on and refine the generated images. Internally, the Agent tries to optimize the policy based on user preferences, ensuring that the final outcomes closely align with user preferences. Various experiments on different tasks demonstrate the effectiveness of the proposed method.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-b9257cbc3aa7426b", "title": "Into the Unknown Unknowns: Engaged Human Learning through Participation in Language Model Agent Conversations", "authors": ["Yucheng Jiang", "Yijia Shao", "Dekun Ma", "Sina J. Semnani", "Monica S. Lam"], "year": 2024, "published": "2024-08-27T17:50:03Z", "updated": "2024-10-17T20:43:22Z", "arxiv_id": "http://arxiv.org/abs/2408.15232v2", "url": "https://arxiv.org/abs/2408.15232v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.IR"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Co-STORM outperforms baseline methods on both discourse trace and report quality."], "abstract": "While language model (LM)-powered chatbots and generative search engines excel at answering concrete queries, discovering information in the terrain of unknown unknowns remains challenging for users. To emulate the common educational scenario where children/students learn by listening to and participating in conversations of their parents/teachers, we create Collaborative STORM (Co-STORM). Unlike QA systems that require users to ask all the questions, Co-STORM lets users observe and occasionally steer the discourse among several LM agents. The agents ask questions on the user's behalf, allowing the user to discover unknown unknowns serendipitously. To facilitate user interaction, Co-STORM assists users in tracking the discourse by organizing the uncovered information into a dynamic mind map, ultimately generating a comprehensive report as takeaways. For automatic evaluation, we construct the WildSeek dataset by collecting real information-seeking records with user goals. Co-STORM outperforms baseline methods on both discourse trace and report quality. In a further human evaluation, 70% of participants prefer Co-STORM over a search engine, and 78% favor it over a RAG chatbot.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-f09aac306d9df482", "title": "Evaluating the Impact of Multiple DER Aggregators on Wholesale Energy Markets: A Hybrid Mean Field Approach", "authors": ["Jun He", "Andrew L. Liu"], "year": 2024, "published": "2024-08-27T14:56:28Z", "updated": "2025-07-04T01:17:58Z", "arxiv_id": "http://arxiv.org/abs/2409.00107v2", "url": "https://arxiv.org/abs/2409.00107v2", "primary_category": "eess.SY", "categories": ["eess.SY", "cs.AI", "cs.LG", "econ.GN", "math.OC"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Multi-Agent", "Quant / Trading"], "benchmarks": ["MATH"], "models": [], "key_results": ["The integration of distributed energy resources (DERs) into wholesale energy markets can greatly enhance grid flexibility, improve market efficiency, and contribute to a more sustainable energy future.", "Furthermore, our results demonstrate that the combination of energy storage and mean-field learning significantly reduces price volatility compared to scenarios without storage."], "abstract": "The integration of distributed energy resources (DERs) into wholesale energy markets can greatly enhance grid flexibility, improve market efficiency, and contribute to a more sustainable energy future. As DERs -- such as solar PV panels and energy storage -- proliferate, effective mechanisms are needed to ensure that small prosumers can participate meaningfully in these markets. We study a wholesale market model featuring multiple DER aggregators, each controlling a portfolio of DER resources and bidding into the market on behalf of the DER asset owners. The key of our approach lies in recognizing the repeated nature of market interactions the ability of participants to learn and adapt over time. Specifically, Aggregators repeatedly interact with each other and with other suppliers in the wholesale market, collectively shaping wholesale electricity prices (aka the locational marginal prices (LMPs)). We model this multi-agent interaction using a mean-field game (MFG), which uses market information -- reflecting the average behavior of market participants -- to enable each aggregator to predict long-term LMP trends and make informed decisions. For each aggregator, because they control the DERs within their portfolio under certain contract structures, we employ a mean-field control (MFC) approach (as opposed to a MFG) to learn an optimal policy that maximizes the total rewards of the DERs under their management. We also propose a reinforcement learning (RL)-based method to help each agent learn optimal strategies within the MFG framework, enhancing their ability to adapt to market conditions and uncertainties. Numerical simulations show that LMPs quickly reach a steady state in the hybrid mean-field approach. Furthermore, our results demonstrate that the combination of energy storage and mean-field learning significantly reduces price volatility compared to scenarios without storage.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-b0863747c95bee75", "title": "No Regrets: Investigating and Improving Regret Approximations for Curriculum Discovery", "authors": ["Alexander Rutherford", "Michael Beukman", "Timon Willi", "Bruno Lacerda", "Nick Hawes", "Jakob Foerster"], "year": 2024, "published": "2024-08-27T14:31:54Z", "updated": "2024-10-29T18:25:44Z", "arxiv_id": "http://arxiv.org/abs/2408.15099v3", "url": "https://arxiv.org/abs/2408.15099v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.RO"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["What data or environments to use for training to improve downstream performance is a longstanding and very topical question in reinforcement learning.", "This simple and intuitive approach outperforms existing UED methods in several binary-outcome environments, including the standard domain of Minigrid and a novel setting closely inspired by a real-world robotics problem."], "abstract": "What data or environments to use for training to improve downstream performance is a longstanding and very topical question in reinforcement learning. In particular, Unsupervised Environment Design (UED) methods have gained recent attention as their adaptive curricula promise to enable agents to be robust to in- and out-of-distribution tasks. This work investigates how existing UED methods select training environments, focusing on task prioritisation metrics. Surprisingly, despite methods aiming to maximise regret in theory, the practical approximations do not correlate with regret but with success rate. As a result, a significant portion of an agent's experience comes from environments it has already mastered, offering little to no contribution toward enhancing its abilities. Put differently, current methods fail to predict intuitive measures of ``learnability.'' Specifically, they are unable to consistently identify those scenarios that the agent can sometimes solve, but not always. Based on our analysis, we develop a method that directly trains on scenarios with high learnability. This simple and intuitive approach outperforms existing UED methods in several binary-outcome environments, including the standard domain of Minigrid and a novel setting closely inspired by a real-world robotics problem. We further introduce a new adversarial evaluation procedure for directly measuring robustness, closely mirroring the conditional value at risk (CVaR). We open-source all our code and present visualisations of final policies here: https://github.com/amacrutherford/sampling-for-learnability.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-40e1c66d3f7a075b", "title": "TourSynbio: A Multi-Modal Large Model and Agent Framework to Bridge Text and Protein Sequences for Protein Engineering", "authors": ["Yiqing Shen", "Zan Chen", "Michail Mamalakis", "Yungeng Liu", "Tianbin Li", "Yanzhou Su", "Junjun He", "Pietro Liò", "Yu Guang Wang"], "year": 2024, "published": "2024-08-27T13:36:00Z", "updated": "2024-08-27T13:36:00Z", "arxiv_id": "http://arxiv.org/abs/2408.15299v1", "url": "https://arxiv.org/abs/2408.15299v1", "primary_category": "q-bio.BM", "categories": ["q-bio.BM", "cs.AI", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["GPT-4"], "key_results": ["While large language models (LLMs) have achieved much progress in the domain of natural language processing, their potential in protein engineering remains largely unexplored.", "TourSynbio-7B outperforms GPT-4 on the ProteinLMBench, a benchmark of 944 manually verified multiple-choice questions, with 62.18% accuracy.", "TourSynbio-Agent integrates previously disconnected deep learning models in the protein engineering domain, offering a unified conversational user interface for improved usability."], "abstract": "The structural similarities between protein sequences and natural languages have led to parallel advancements in deep learning across both domains. While large language models (LLMs) have achieved much progress in the domain of natural language processing, their potential in protein engineering remains largely unexplored. Previous approaches have equipped LLMs with protein understanding capabilities by incorporating external protein encoders, but this fails to fully leverage the inherent similarities between protein sequences and natural languages, resulting in sub-optimal performance and increased model complexity. To address this gap, we present TourSynbio-7B, the first multi-modal large model specifically designed for protein engineering tasks without external protein encoders. TourSynbio-7B demonstrates that LLMs can inherently learn to understand proteins as language. The model is post-trained and instruction fine-tuned on InternLM2-7B using ProteinLMDataset, a dataset comprising 17.46 billion tokens of text and protein sequence for self-supervised pretraining and 893K instructions for supervised fine-tuning. TourSynbio-7B outperforms GPT-4 on the ProteinLMBench, a benchmark of 944 manually verified multiple-choice questions, with 62.18% accuracy. Leveraging TourSynbio-7B's enhanced protein sequence understanding capability, we introduce TourSynbio-Agent, an innovative framework capable of performing various protein engineering tasks, including mutation analysis, inverse folding, protein folding, and visualization. TourSynbio-Agent integrates previously disconnected deep learning models in the protein engineering domain, offering a unified conversational user interface for improved usability. Finally, we demonstrate the efficacy of TourSynbio-7B and TourSynbio-Agent through two wet lab case studies on vanilla key enzyme modification and steroid compound catalysis.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-720ba64ec4016ecf", "title": "AgentMonitor: A Plug-and-Play Framework for Predictive and Secure Multi-Agent Systems", "authors": ["Chi-Min Chan", "Jianxuan Yu", "Weize Chen", "Chunyang Jiang", "Xinyu Liu", "Weijie Shi", "Zhiyuan Liu", "Wei Xue", "Yike Guo"], "year": 2024, "published": "2024-08-27T11:24:38Z", "updated": "2024-08-27T11:24:38Z", "arxiv_id": "http://arxiv.org/abs/2408.14972v1", "url": "https://arxiv.org/abs/2408.14972v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Safety / Jailbreak", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Recent research shows that multi-agent systems (MAS), where each agent plays a specific role, can outperform individual LLMs.", "Experiments demonstrate that an XGBoost model achieves a Spearman correlation of 0.89 in-domain and 0.58 in more challenging scenarios.", "Furthermore, using AgentMonitor reduces harmful content by 6.2% and increases helpful content by 1.8% on average, enhancing safety and reliability."], "abstract": "The rapid advancement of large language models (LLMs) has led to the rise of LLM-based agents. Recent research shows that multi-agent systems (MAS), where each agent plays a specific role, can outperform individual LLMs. However, configuring an MAS for a task remains challenging, with performance only observable post-execution. Inspired by scaling laws in LLM development, we investigate whether MAS performance can be predicted beforehand. We introduce AgentMonitor, a framework that integrates at the agent level to capture inputs and outputs, transforming them into statistics for training a regression model to predict task performance. Additionally, it can further apply real-time corrections to address security risks posed by malicious agents, mitigating negative impacts and enhancing MAS security. Experiments demonstrate that an XGBoost model achieves a Spearman correlation of 0.89 in-domain and 0.58 in more challenging scenarios. Furthermore, using AgentMonitor reduces harmful content by 6.2% and increases helpful content by 1.8% on average, enhancing safety and reliability. Code is available at \\url{https://github.com/chanchimin/AgentMonitor}.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-14b5c37c6718e2ec", "title": "Decentralized Unlabeled Multi-agent Pathfinding Via Target And Priority Swapping (With Supplementary)", "authors": ["Stepan Dergachev", "Konstantin Yakovlev"], "year": 2024, "published": "2024-08-27T10:45:57Z", "updated": "2024-08-27T10:45:57Z", "arxiv_id": "http://arxiv.org/abs/2408.14948v1", "url": "https://arxiv.org/abs/2408.14948v1", "primary_category": "cs.MA", "categories": ["cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["Indeed, TP-SWAP outperforms the fully-decentralized competitor and can even outperform the semi-decentralized one (i.e."], "abstract": "In this paper we study a challenging variant of the multi-agent pathfinding problem (MAPF), when a set of agents must reach a set of goal locations, but it does not matter which agent reaches a specific goal - Anonymous MAPF (AMAPF). Current optimal and suboptimal AMAPF solvers rely on the existence of a centralized controller which is in charge of both target assignment and pathfinding. We extend the state of the art and present the first AMAPF solver capable of solving the problem at hand in a fully decentralized fashion, when each agent makes decisions individually and relies only on the local communication with the others. The core of our method is a priority and target swapping procedure tailored to produce consistent goal assignments (i.e. making sure that no two agents are heading towards the same goal). Coupled with an established rule-based path planning, we end up with a TP-SWAP, an efficient and flexible approach to solve decentralized AMAPF. On the theoretical side, we prove that TP-SWAP is complete (i.e. TP-SWAP guarantees that each target will be reached by some agent). Empirically, we evaluate TP-SWAP across a wide range of setups and compare it to both centralized and decentralized baselines. Indeed, TP-SWAP outperforms the fully-decentralized competitor and can even outperform the semi-decentralized one (i.e. the one relying on the initial consistent goal assignment) in terms of flowtime (a widespread cost objective in MAPF", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-fa8e4f2208ffc197", "title": "Learning Robust Reward Machines from Noisy Labels", "authors": ["Roko Parac", "Lorenzo Nodari", "Leo Ardon", "Daniel Furelos-Blanco", "Federico Cerutti", "Alessandra Russo"], "year": 2024, "published": "2024-08-27T08:41:42Z", "updated": "2025-03-21T14:07:55Z", "arxiv_id": "http://arxiv.org/abs/2408.14871v2", "url": "https://arxiv.org/abs/2408.14871v2", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["PROB-IRM uses a state-of-the-art inductive logic programming framework robust to noisy examples to learn RMs from noisy traces using the Bayesian posterior degree of beliefs, thus ensuring robustness against inconsistencies."], "abstract": "This paper presents PROB-IRM, an approach that learns robust reward machines (RMs) for reinforcement learning (RL) agents from noisy execution traces. The key aspect of RM-driven RL is the exploitation of a finite-state machine that decomposes the agent's task into different subtasks. PROB-IRM uses a state-of-the-art inductive logic programming framework robust to noisy examples to learn RMs from noisy traces using the Bayesian posterior degree of beliefs, thus ensuring robustness against inconsistencies. Pivotal for the results is the interleaving between RM learning and policy learning: a new RM is learned whenever the RL agent generates a trace that is believed not to be accepted by the current RM. To speed up the training of the RL agent, PROB-IRM employs a probabilistic formulation of reward shaping that uses the posterior Bayesian beliefs derived from the traces. Our experimental analysis shows that PROB-IRM can learn (potentially imperfect) RMs from noisy traces and exploit them to train an RL agent to solve its tasks successfully. Despite the complexity of learning the RM from noisy traces, agents trained with PROB-IRM perform comparably to agents provided with handcrafted RMs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-567dc7cc9552b80f", "title": "MA-CDMR: An Intelligent Cross-domain Multicast Routing Method based on Multiagent Deep Reinforcement Learning in Multi-domain SDWN", "authors": ["Miao Ye", "Hongwen Hu", "Xiaoli Wang", "Yuping Wang", "Yong Wang", "Wen Peng", "Jihao Zheng"], "year": 2024, "published": "2024-08-27T08:16:32Z", "updated": "2024-09-11T13:52:05Z", "arxiv_id": "http://arxiv.org/abs/2409.05888v2", "url": "https://arxiv.org/abs/2409.05888v2", "primary_category": "cs.NI", "categories": ["cs.NI", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["The cross-domain multicast routing problem in a software-defined wireless network with multiple controllers is a classic NP-hard optimization problem."], "abstract": "The cross-domain multicast routing problem in a software-defined wireless network with multiple controllers is a classic NP-hard optimization problem. As the network size increases, designing and implementing cross-domain multicast routing paths in the network requires not only designing efficient solution algorithms to obtain the optimal cross-domain multicast tree but also ensuring the timely and flexible acquisition and maintenance of global network state information. However, existing solutions have a limited ability to sense the network traffic state, affecting the quality of service of multicast services. In addition, these methods have difficulty adapting to the highly dynamically changing network states and have slow convergence speeds. To this end, this paper aims to design and implement a multiagent deep reinforcement learning based cross-domain multicast routing method for SDWN with multicontroller domains. First, a multicontroller communication mechanism and a multicast group management module are designed to transfer and synchronize network information between different control domains of the SDWN, thus effectively managing the joining and classification of members in the cross-domain multicast group. Second, a theoretical analysis and proof show that the optimal cross-domain multicast tree includes an interdomain multicast tree and an intradomain multicast tree. An agent is established for each controller, and a cooperation mechanism between multiple agents is designed to effectively optimize cross-domain multicast routing and ensure consistency and validity in the representation of network state information for cross-domain multicast routing decisions. Third, a multiagent reinforcement learning-based method that combines online and offline training is designed to reduce the dependence on the real-time environment and increase the convergence speed of multiple agents.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-26e6d6899504823a", "title": "Diffusion Models Are Real-Time Game Engines", "authors": ["Dani Valevski", "Yaniv Leviathan", "Moab Arar", "Shlomi Fruchter"], "year": 2024, "published": "2024-08-27T07:46:07Z", "updated": "2025-04-24T03:03:57Z", "arxiv_id": "http://arxiv.org/abs/2408.14837v2", "url": "https://arxiv.org/abs/2408.14837v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CV"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Next frame prediction achieves a PSNR of 29.4, comparable to lossy JPEG compression.", "Conditioning augmentations help ensure stable auto-regressive generation over long trajectories, and decoder fine-tuning improves the fidelity of visual details and text."], "abstract": "We present GameNGen, the first game engine powered entirely by a neural model that also enables real-time interaction with a complex environment over long trajectories at high quality. When trained on the classic game DOOM, GameNGen extracts gameplay and uses it to generate a playable environment that can interactively simulate new trajectories. GameNGen runs at 20 frames per second on a single TPU and remains stable over extended multi-minute play sessions. Next frame prediction achieves a PSNR of 29.4, comparable to lossy JPEG compression. Human raters are only slightly better than random chance at distinguishing short clips of the game from clips of the simulation, even after 5 minutes of auto-regressive generation. GameNGen is trained in two phases: (1) an RL-agent learns to play the game and the training sessions are recorded, and (2) a diffusion model is trained to produce the next frame, conditioned on the sequence of past frames and actions. Conditioning augmentations help ensure stable auto-regressive generation over long trajectories, and decoder fine-tuning improves the fidelity of visual details and text.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-852daf56d0a74090", "title": "Sequential-Scanning Dual-Energy CT Imaging Using High Temporal Resolution Image Reconstruction and Error-Compensated Material Basis Image Generation", "authors": ["Qiaoxin Li", "Ruifeng Chen", "Peng Wang", "Guotao Quan", "Yanfeng Du", "Dong Liang", "Yinsheng Li"], "year": 2024, "published": "2024-08-27T03:09:39Z", "updated": "2024-08-27T03:09:39Z", "arxiv_id": "http://arxiv.org/abs/2408.14754v1", "url": "https://arxiv.org/abs/2408.14754v1", "primary_category": "physics.med-ph", "categories": ["physics.med-ph", "cs.AI", "cs.CV", "physics.ins-det"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["In this work, we developed sequential-scanning DECT imaging using high temporal resolution image reconstruction and error-compensated material basis image generation, ACCELERATION in short, to address the technical challenge induced by temporal inconsistency of sequentially scanned data sets and improve quantification accuracy of material concentration in sequential-scanning DECT.", "Results demonstrated the improvement of quantification accuracy and image quality using ACCELERATION."], "abstract": "Dual-energy computed tomography (DECT) has been widely used to obtain quantitative elemental composition of imaged subjects for personalized and precise medical diagnosis. Compared with DECT leveraging advanced X-ray source and/or detector technologies, the use of the sequential-scanning data acquisition scheme to implement DECT may make a broader impact on clinical practice because this scheme requires no specialized hardware designs and can be directly implemented into conventional CT systems. However, since the concentration of iodinated contrast agent in the imaged subject varies over time, sequentially scanned data sets acquired at two tube potentials are temporally inconsistent. As existing material basis image reconstruction approaches assume that the data sets acquired at two tube potentials are temporally consistent, the violation of this assumption results in inaccurate quantification of material concentration. In this work, we developed sequential-scanning DECT imaging using high temporal resolution image reconstruction and error-compensated material basis image generation, ACCELERATION in short, to address the technical challenge induced by temporal inconsistency of sequentially scanned data sets and improve quantification accuracy of material concentration in sequential-scanning DECT. ACCELERATION has been validated and evaluated using numerical simulation data sets generated from clinical human subject exams and experimental human subject studies. Results demonstrated the improvement of quantification accuracy and image quality using ACCELERATION.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4cad9ec1591a1e0d", "title": "Bidirectional Emergent Language in Situated Environments", "authors": ["Cornelius Wolff", "Julius Mayer", "Elia Bruni", "Xenia Ohmer"], "year": 2024, "published": "2024-08-26T21:25:44Z", "updated": "2024-10-17T10:55:35Z", "arxiv_id": "http://arxiv.org/abs/2408.14649v2", "url": "https://arxiv.org/abs/2408.14649v2", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["These environments are interesting because optimal performance requires the emergence of a communication protocol, but moderate success can be achieved without one."], "abstract": "Emergent language research has made significant progress in recent years, but still largely fails to explore how communication emerges in more complex and situated multi-agent systems. Existing setups often employ a reference game, which limits the range of language emergence phenomena that can be studied, as the game consists of a single, purely language-based interaction between the agents. In this paper, we address these limitations and explore the emergence and utility of token-based communication in open-ended multi-agent environments, where situated agents interact with the environment through movement and communication over multiple time-steps. Specifically, we introduce two novel cooperative environments: Multi-Agent Pong and Collectors. These environments are interesting because optimal performance requires the emergence of a communication protocol, but moderate success can be achieved without one. By employing various methods from explainable AI research, such as saliency maps, perturbation, and diagnostic classifiers, we are able to track and interpret the agents' language channel use over time. We find that the emerging communication is sparse, with the agents only generating meaningful messages and acting upon incoming messages in states where they cannot succeed without coordination.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-34bc29b10a763e16", "title": "On Centralized Critics in Multi-Agent Reinforcement Learning", "authors": ["Xueguang Lyu", "Andrea Baisero", "Yuchen Xiao", "Brett Daley", "Christopher Amato"], "year": 2024, "published": "2024-08-26T19:27:06Z", "updated": "2024-08-26T19:27:06Z", "arxiv_id": "http://arxiv.org/abs/2408.14597v1", "url": "https://arxiv.org/abs/2408.14597v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Safety / Jailbreak", "Multi-Agent", "Deep Research"], "benchmarks": [], "models": [], "key_results": ["Centralized Training for Decentralized Execution where agents are trained offline in a centralized fashion and execute online in a decentralized manner, has become a popular approach in Multi-Agent Reinforcement Learning (MARL)."], "abstract": "Centralized Training for Decentralized Execution where agents are trained offline in a centralized fashion and execute online in a decentralized manner, has become a popular approach in Multi-Agent Reinforcement Learning (MARL). In particular, it has become popular to develop actor-critic methods that train decentralized actors with a centralized critic where the centralized critic is allowed access global information of the entire system, including the true system state. Such centralized critics are possible given offline information and are not used for online execution. While these methods perform well in a number of domains and have become a de facto standard in MARL, using a centralized critic in this context has yet to be sufficiently analyzed theoretically or empirically. In this paper, we therefore formally analyze centralized and decentralized critic approaches, and analyze the effect of using state-based critics in partially observable environments. We derive theories contrary to the common intuition: critic centralization is not strictly beneficial, and using state values can be harmful. We further prove that, in particular, state-based critics can introduce unexpected bias and variance compared to history-based critics. Finally, we demonstrate how the theory applies in practice by comparing different forms of critics on a wide range of common multi-agent benchmarks. The experiments show practical issues such as the difficulty of representation learning with partial observability, which highlights why the theoretical problems are often overlooked in the literature.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-49776a1109292add", "title": "Students' Perceived Roles, Opportunities, and Challenges of a Generative AI-powered Teachable Agent: A Case of Middle School Math Class", "authors": ["Yukyeong Song", "Jinhee Kim", "Zifeng Liu", "Chenglu Li", "Wanli Xing"], "year": 2024, "published": "2024-08-26T18:54:20Z", "updated": "2024-08-26T18:54:20Z", "arxiv_id": "http://arxiv.org/abs/2409.06721v1", "url": "https://arxiv.org/abs/2409.06721v1", "primary_category": "cs.CY", "categories": ["cs.CY", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["Ongoing advancements in Generative AI (GenAI) have boosted the potential of applying long-standing learning-by-teaching practices in the form of a teachable agent (TA)."], "abstract": "Ongoing advancements in Generative AI (GenAI) have boosted the potential of applying long-standing learning-by-teaching practices in the form of a teachable agent (TA). Despite the recognized roles and opportunities of TAs, less is known about how GenAI could create synergy or introduce challenges in TAs and how students perceived the application of GenAI in TAs. This study explored middle school students perceived roles, benefits, and challenges of GenAI-powered TAs in an authentic mathematics classroom. Through classroom observation, focus-group interviews, and open-ended surveys of 108 sixth-grade students, we found that students expected the GenAI-powered TA to serve as a learning companion, facilitator, and collaborative problem-solver. Students also expressed the benefits and challenges of GenAI-powered TAs. This study provides implications for the design of educational AI and AI-assisted instruction.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4490e00116d2c8af", "title": "EVINCE: Optimizing Multi-LLM Dialogues Using Conditional Statistics and Information Theory", "authors": ["Edward Y. Chang"], "year": 2024, "published": "2024-08-26T18:48:51Z", "updated": "2025-01-29T20:48:59Z", "arxiv_id": "http://arxiv.org/abs/2408.14575v4", "url": "https://arxiv.org/abs/2408.14575v4", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["EVINCE (Entropy and Variation IN Conditional Exchanges) is a novel framework for optimizing multi-LLM dialogues using conditional statistics and information theory."], "abstract": "EVINCE (Entropy and Variation IN Conditional Exchanges) is a novel framework for optimizing multi-LLM dialogues using conditional statistics and information theory. It addresses limitations in multi-agent debate (MAS) frameworks, where multiple LLMs ``chat'' without behavior modulation or mutual information quality assessment. Using dual entropy optimization to balance perspective diversity and prior knowledge, $\\EVINCE$ provides quantitative tools to dynamically regulate LLM linguistic behaviors. When mutual information is low and both cross-entropy and Wasserstein distance are high, EVINCE promotes contentious dialogues to expose diverse perspectives and uncover inconsistencies. Conversely, as cross-entropy decreases and mutual information stabilizes, it transitions discussions into a conciliatory phase, encouraging compromise and acknowledgment of valid points. Using information-theoretic metrics and optimizing mutual information, $\\EVINCE$ emerges as a structured and highly effective framework for multi-LLM collaboration.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-6090ee505ca550b1", "title": "Squid: Long Context as a New Modality for Energy-Efficient On-Device Language Models", "authors": ["Wei Chen", "Zhiyuan Li", "Shuo Xin", "Yihao Wang"], "year": 2024, "published": "2024-08-28T04:06:14Z", "updated": "2026-07-20T02:08:41Z", "arxiv_id": "http://arxiv.org/abs/2408.15518v3", "url": "https://arxiv.org/abs/2408.15518v3", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Long Context", "Memory / RAG", "VLM"], "benchmarks": [], "models": ["phin", "phisticated", "phin."], "key_results": ["Empirical evaluations demonstrate a 10-fold improvement in energy efficiency and a 5-fold reduction in latency compared to conventional full-length context processing methods without losing quality of the response."], "abstract": "This paper presents Dolphin, a novel decoder-decoder architecture for energy-efficient processing of long contexts in language models. Our approach addresses the significant energy consumption and latency challenges inherent in on-device models. Dolphin employs a compact 0.5B parameter decoder to distill extensive contextual information into a memory embedding, substantially reducing the input length for the primary 7B parameter decoder model. Inspired by vision-language models, we repurpose the image embedding projector to encode long textual contexts, effectively treating extended context as a distinct modality. This innovative method enables processing of substantially longer contexts without the typical computational overhead associated with extended input sequences. Empirical evaluations demonstrate a 10-fold improvement in energy efficiency and a 5-fold reduction in latency compared to conventional full-length context processing methods without losing quality of the response. Our work contributes to the development of more sustainable and scalable language models for on-device applications, addressing the critical need for energy-efficient and responsive AI technologies in resource-constrained environments while maintaining the accuracy to understand long contexts. This research has implications for the broader field of natural language processing, particularly in the domain of efficient model design for resource-limited settings. By enabling more sophisticated AI capabilities on edge devices, Dolphin paves the way for advanced language processing in a wide range of applications where computational resources are at a premium. The Dolphin model is publicly available at https://huggingface.co/NexaAIDev/Dolphin.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-01beb85610c70146", "title": "ReMamba: Equip Mamba with Effective Long-Sequence Modeling", "authors": ["Danlong Yuan", "Jiahao Liu", "Bei Li", "Huishuai Zhang", "Jingang Wang", "Xunliang Cai", "Dongyan Zhao"], "year": 2024, "published": "2024-08-28T02:47:27Z", "updated": "2025-01-01T15:22:34Z", "arxiv_id": "http://arxiv.org/abs/2408.15496v4", "url": "https://arxiv.org/abs/2408.15496v4", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Long Context"], "benchmarks": [], "models": [], "key_results": ["While the Mamba architecture demonstrates superior inference efficiency and competitive performance on short-context natural language processing (NLP) tasks, empirical evidence suggests its capacity to comprehend long contexts is limited compared to transformer-based models."], "abstract": "While the Mamba architecture demonstrates superior inference efficiency and competitive performance on short-context natural language processing (NLP) tasks, empirical evidence suggests its capacity to comprehend long contexts is limited compared to transformer-based models. In this study, we investigate the long-context efficiency issues of the Mamba models and propose ReMamba, which enhances Mamba's ability to comprehend long contexts. ReMamba incorporates selective compression and adaptation techniques within a two-stage re-forward process, incurring minimal additional inference costs overhead. Experimental results on the LongBench and L-Eval benchmarks demonstrate ReMamba's efficacy, improving over the baselines by 3.2 and 1.6 points, respectively, and attaining performance almost on par with same-size transformer models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-a4a49ec1d5326a4b", "title": "The Mamba in the Llama: Distilling and Accelerating Hybrid Models", "authors": ["Junxiong Wang", "Daniele Paliotta", "Avner May", "Alexander M. Rush", "Tri Dao"], "year": 2024, "published": "2024-08-27T17:56:11Z", "updated": "2025-06-27T07:54:57Z", "arxiv_id": "http://arxiv.org/abs/2408.15237v4", "url": "https://arxiv.org/abs/2408.15237v4", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Speculative Decoding"], "benchmarks": ["AlpacaEval", "MT-Bench"], "models": ["GPT-4", "Llama", "Llama3-8B-Instruct", "Llama and"], "key_results": ["The resulting hybrid model, which incorporates a quarter of the attention layers, achieves performance comparable to the original Transformer in chat benchmarks and outperforms open-source hybrid Mamba models trained from scratch with trillions of tokens in both chat benchmarks and general benchmarks.", "Our top-performing model, distilled from Llama3-8B-Instruct, achieves a 29.61 length-controlled win rate on AlpacaEval 2 against GPT-4 and 7.35 on MT-Bench, surpassing the best 8B scale instruction-tuned linear RNN model."], "abstract": "Linear RNN architectures, like Mamba, can be competitive with Transformer models in language modeling while having advantageous deployment characteristics. Given the focus on training large-scale Transformer models, we consider the challenge of converting these pretrained models for deployment. We demonstrate that it is feasible to distill large Transformers into linear RNNs by reusing the linear projection weights from attention layers with academic GPU resources. The resulting hybrid model, which incorporates a quarter of the attention layers, achieves performance comparable to the original Transformer in chat benchmarks and outperforms open-source hybrid Mamba models trained from scratch with trillions of tokens in both chat benchmarks and general benchmarks. Moreover, we introduce a hardware-aware speculative decoding algorithm that accelerates the inference speed of Mamba and hybrid models. Overall we show how, with limited computation resources, we can remove many of the original attention layers and generate from the resulting model more efficiently. Our top-performing model, distilled from Llama3-8B-Instruct, achieves a 29.61 length-controlled win rate on AlpacaEval 2 against GPT-4 and 7.35 on MT-Bench, surpassing the best 8B scale instruction-tuned linear RNN model. We also find that the distilled model has natural length extrapolation, showing almost perfect accuracy in the needle-in-a-haystack test at 20x the distillation length. Code and pre-trained checkpoints are open-sourced at https://github.com/jxiw/MambaInLlama and https://github.com/itsdaniele/speculative_mamba.", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-e65050f1941ece10", "title": "Parameter-Efficient Quantized Mixture-of-Experts Meets Vision-Language Instruction Tuning for Semiconductor Electron Micrograph Analysis", "authors": ["Sakhinana Sagar Srinivas", "Chidaksh Ravuru", "Geethan Sannidhi", "Venkataramana Runkana"], "year": 2024, "published": "2024-08-27T15:59:26Z", "updated": "2024-08-27T15:59:26Z", "arxiv_id": "http://arxiv.org/abs/2408.15305v1", "url": "https://arxiv.org/abs/2408.15305v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "Quantization", "Computer Use", "VLM", "Quant / Trading"], "benchmarks": [], "models": ["GPT-4"], "key_results": ["Rigorous experiments validate that our framework surpasses traditional methods, handles data shifts, and enables high-throughput screening."], "abstract": "Semiconductors, crucial to modern electronics, are generally under-researched in foundational models. It highlights the need for research to enhance the semiconductor device technology portfolio and aid in high-end device fabrication. In this paper, we introduce sLAVA, a small-scale vision-language assistant tailored for semiconductor manufacturing, with a focus on electron microscopy image analysis. It addresses challenges of data scarcity and acquiring high-quality, expert-annotated data. We employ a teacher-student paradigm, using a foundational vision language model like GPT-4 as a teacher to create instruction-following multimodal data for customizing the student model, sLAVA, for electron microscopic image analysis tasks on consumer hardware with limited budgets. Our approach allows enterprises to further fine-tune the proposed framework with their proprietary data securely within their own infrastructure, protecting intellectual property. Rigorous experiments validate that our framework surpasses traditional methods, handles data shifts, and enables high-throughput screening.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-d4f0f81491b32c2a", "title": "The Uniqueness of LLaMA3-70B Series with Per-Channel Quantization", "authors": ["Minghai Qin"], "year": 2024, "published": "2024-08-27T15:03:01Z", "updated": "2024-10-01T09:05:45Z", "arxiv_id": "http://arxiv.org/abs/2408.15301v2", "url": "https://arxiv.org/abs/2408.15301v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["LLaMA3-70B", "LLaMA3", "LLaMA2-70B", "LLaMA2", "LLaMA3.2", "Qwen", "Mistral", "Mixtral", "Phi-3", "Falcon demonstrate robust performance with W8A8. Contrary to previous assertions attributing degradation to the large dynamic range of activations", "ying"], "key_results": ["Experimental results demonstrate that both strategies effectively preserve the accuracy of the entire LLaMA3-70B model series under W8A8 quantization, achieving performance on par with their FP16 counterparts."], "abstract": "We have observed a distinctive quantization-related behavior in the LLaMA3/3.1-70B models that is absent in both the LLaMA2-70B and LLaMA3/3.1/3.2-1B/3B/8B/405B models. Quantization is a crucial technique for deploying large language models (LLMs) efficiently. The impact of W8A8 post-training quantization on model accuracy, especially on the recently released LLaMA3/3.1 model series, remains contentious. In this paper, we explore three key questions: What makes the LLaMA3-70B model series uniquely vulnerable to quantization? Why is this the case? And how can the issue be addressed? We empirically investigate multiple LLMs featured on an open LLM leaderboard, discovering that the LLaMA3-70B model series have a unique accuracy degradation behavior with W8A8 per-channel post-training quantization. In contrast, other model series such as LLaMA2, LLaMA3/3.1-8B, LLaMA3.2, Qwen, Mixtral, Mistral, Phi-3, and Falcon demonstrate robust performance with W8A8. Contrary to previous assertions attributing degradation to the large dynamic range of activations, our findings indicate that the weight distribution of the LLaMA3-70B is the primary factor behind the vulnerability. By meticulously analyzing the distinct characteristics of weight distributions across Transformer blocks, we propose two solutions that make different tradeoffs in hardware/software overhead. First, we propose a mixed strategy where less than 3\\% of the layers employ finer per-group W8A8 quantization granularity. Second, we introduce a bi-smoothing strategy that balances quantization errors between weights and activations while maintaining per-channel quantization throughout. Experimental results demonstrate that both strategies effectively preserve the accuracy of the entire LLaMA3-70B model series under W8A8 quantization, achieving performance on par with their FP16 counterparts.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-3955a77e03c49f80", "title": "GIFT-SW: Gaussian noise Injected Fine-Tuning of Salient Weights for LLMs", "authors": ["Maxim Zhelnin", "Viktor Moskvoretskii", "Egor Shvetsov", "Egor Venediktov", "Mariya Krylova", "Aleksandr Zuev", "Evgeny Burnaev"], "year": 2024, "published": "2024-08-27T14:41:14Z", "updated": "2024-08-27T14:41:14Z", "arxiv_id": "http://arxiv.org/abs/2408.15300v1", "url": "https://arxiv.org/abs/2408.15300v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": ["LLaMA models"], "key_results": ["Experiments with LLaMA models demonstrate that GIFT-SW outperforms full fine-tuning and modern PEFT methods under the same computational budget."], "abstract": "Parameter Efficient Fine-Tuning (PEFT) methods have gained popularity and democratized the usage of Large Language Models (LLMs). Recent studies have shown that a small subset of weights significantly impacts performance. Based on this observation, we introduce a novel PEFT method, called Gaussian noise Injected Fine Tuning of Salient Weights (GIFT-SW). Our method updates only salient columns, while injecting Gaussian noise into non-salient ones. To identify these columns, we developeda generalized sensitivity metric that extends and unifies metrics from previous studies. Experiments with LLaMA models demonstrate that GIFT-SW outperforms full fine-tuning and modern PEFT methods under the same computational budget. Moreover, GIFT-SW offers practical advantages to recover performance of models subjected to mixed-precision quantization with keeping salient weights in full precision.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4d8e9843fbf5bb7d", "title": "Evidence-Enhanced Triplet Generation Framework for Hallucination Alleviation in Generative Question Answering", "authors": ["Haowei Du", "Huishuai Zhang", "Dongyan Zhao"], "year": 2024, "published": "2024-08-27T13:07:07Z", "updated": "2024-08-27T13:07:07Z", "arxiv_id": "http://arxiv.org/abs/2408.15037v1", "url": "https://arxiv.org/abs/2408.15037v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["LLama and"], "key_results": ["Our framework ensures the model to learn the logical relation between query, evidence and answer, which simultaneously improves the evidence generation and query answering.", "In this paper, we apply EATQA to LLama and it outperforms other LLMs-based methods and hallucination mitigation approaches on two challenging GQA benchmarks."], "abstract": "To address the hallucination in generative question answering (GQA) where the answer can not be derived from the document, we propose a novel evidence-enhanced triplet generation framework, EATQA, encouraging the model to predict all the combinations of (Question, Evidence, Answer) triplet by flipping the source pair and the target label to understand their logical relationships, i.e., predict Answer(A), Question(Q), and Evidence(E) given a QE, EA, and QA pairs, respectively. Furthermore, we bridge the distribution gap to distill the knowledge from evidence in inference stage. Our framework ensures the model to learn the logical relation between query, evidence and answer, which simultaneously improves the evidence generation and query answering. In this paper, we apply EATQA to LLama and it outperforms other LLMs-based methods and hallucination mitigation approaches on two challenging GQA benchmarks. Further analysis shows that our method not only keeps prior knowledge within LLM, but also mitigates hallucination and generates faithful answers.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7ea39c7435411991", "title": "NeuroLM: A Universal Multi-task Foundation Model for Bridging the Gap between Language and EEG Signals", "authors": ["Wei-Bang Jiang", "Yansen Wang", "Bao-Liang Lu", "Dongsheng Li"], "year": 2024, "published": "2024-08-27T12:07:09Z", "updated": "2025-03-20T08:26:21Z", "arxiv_id": "http://arxiv.org/abs/2409.00101v3", "url": "https://arxiv.org/abs/2409.00101v3", "primary_category": "eess.SP", "categories": ["eess.SP", "cs.HC", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["Recent advancements for large-scale pre-training with neural signals such as electroencephalogram (EEG) have shown promising results, significantly boosting the development of brain-computer interfaces (BCIs) and healthcare.", "However, these pre-trained models often require full fine-tuning on each downstream task to achieve substantial improvements, limiting their versatility and usability, and leading to considerable resource wastage."], "abstract": "Recent advancements for large-scale pre-training with neural signals such as electroencephalogram (EEG) have shown promising results, significantly boosting the development of brain-computer interfaces (BCIs) and healthcare. However, these pre-trained models often require full fine-tuning on each downstream task to achieve substantial improvements, limiting their versatility and usability, and leading to considerable resource wastage. To tackle these challenges, we propose NeuroLM, the first multi-task foundation model that leverages the capabilities of Large Language Models (LLMs) by regarding EEG signals as a foreign language, endowing the model with multi-task learning and inference capabilities. Our approach begins with learning a text-aligned neural tokenizer through vector-quantized temporal-frequency prediction, which encodes EEG signals into discrete neural tokens. These EEG tokens, generated by the frozen vector-quantized (VQ) encoder, are then fed into an LLM that learns causal EEG information via multi-channel autoregression. Consequently, NeuroLM can understand both EEG and language modalities. Finally, multi-task instruction tuning adapts NeuroLM to various downstream tasks. We are the first to demonstrate that, by specific incorporation with LLMs, NeuroLM unifies diverse EEG tasks within a single model through instruction tuning. The largest variant NeuroLM-XL has record-breaking 1.7B parameters for EEG signal processing, and is pre-trained on a large-scale corpus comprising approximately 25,000-hour EEG data. When evaluated on six diverse downstream datasets, NeuroLM showcases the huge potential of this multi-task learning paradigm.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-1c5f2632a73a079e", "title": "MRSE: An Efficient Multi-modality Retrieval System for Large Scale E-commerce", "authors": ["Hao Jiang", "Haoxiang Zhang", "Qingshan Hou", "Chaofeng Chen", "Weisi Lin", "Jingchang Zhang", "Annan Wang"], "year": 2024, "published": "2024-08-27T11:21:19Z", "updated": "2024-08-27T11:21:19Z", "arxiv_id": "http://arxiv.org/abs/2408.14968v1", "url": "https://arxiv.org/abs/2408.14968v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": [], "benchmarks": [], "models": [], "key_results": ["Experiments on a large-scale dataset from Shopee and online A/B testing show that MRSE achieves an 18.9% improvement in offline relevance and a 3.7% gain in online core metrics compared to Shopee's state-of-the-art uni-modality system."], "abstract": "Providing high-quality item recall for text queries is crucial in large-scale e-commerce search systems. Current Embedding-based Retrieval Systems (ERS) embed queries and items into a shared low-dimensional space, but uni-modality ERS rely too heavily on textual features, making them unreliable in complex contexts. While multi-modality ERS incorporate various data sources, they often overlook individual preferences for different modalities, leading to suboptimal results. To address these issues, we propose MRSE, a Multi-modality Retrieval System that integrates text, item images, and user preferences through lightweight mixture-of-expert (LMoE) modules to better align features across and within modalities. MRSE also builds user profiles at a multi-modality level and introduces a novel hybrid loss function that enhances consistency and robustness using hard negative sampling. Experiments on a large-scale dataset from Shopee and online A/B testing show that MRSE achieves an 18.9% improvement in offline relevance and a 3.7% gain in online core metrics compared to Shopee's state-of-the-art uni-modality system.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-874b260b6c5f1718", "title": "Writing in the Margins: Better Inference Pattern for Long Context Retrieval", "authors": ["Melisa Russak", "Umar Jamil", "Christopher Bryant", "Kiran Kamble", "Axel Magnuson", "Mateusz Russak", "Waseem AlShikh"], "year": 2024, "published": "2024-08-27T09:34:38Z", "updated": "2024-08-27T09:34:38Z", "arxiv_id": "http://arxiv.org/abs/2408.14906v1", "url": "https://arxiv.org/abs/2408.14906v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.IR"], "pillar": "Efficient Training & Inference", "methods": ["Long Context", "Memory / RAG"], "benchmarks": ["HotpotQA"], "models": [], "key_results": ["In this paper, we introduce Writing in the Margins (WiM), a new inference pattern for Large Language Models designed to optimize the handling of long input sequences in retrieval-oriented tasks."], "abstract": "In this paper, we introduce Writing in the Margins (WiM), a new inference pattern for Large Language Models designed to optimize the handling of long input sequences in retrieval-oriented tasks. This approach leverages the chunked prefill of the key-value cache to perform segment-wise inference, which enables efficient processing of extensive contexts along with the generation and classification of intermediate information (\"margins\") that guide the model towards specific tasks. This method increases computational overhead marginally while significantly enhancing the performance of off-the-shelf models without the need for fine-tuning. Specifically, we observe that WiM provides an average enhancement of 7.5% in accuracy for reasoning skills (HotpotQA, MultiHop-RAG) and more than a 30.0% increase in the F1-score for aggregation tasks (CWE). Additionally, we show how the proposed pattern fits into an interactive retrieval design that provides end-users with ongoing updates about the progress of context processing, and pinpoints the integration of relevant information into the final response. We release our implementation of WiM using Hugging Face Transformers library at https://github.com/writer/writing-in-the-margins.", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-5f083bbae063fc3f", "title": "Learning Differentially Private Diffusion Models via Stochastic Adversarial Distillation", "authors": ["Bochao Liu", "Pengju Wang", "Shiming Ge"], "year": 2024, "published": "2024-08-27T02:29:29Z", "updated": "2024-08-27T02:29:29Z", "arxiv_id": "http://arxiv.org/abs/2408.14738v1", "url": "https://arxiv.org/abs/2408.14738v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CR", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Specifically, we first train a diffusion model as a teacher and then train a student by distillation, in which we achieve differential privacy by adding noise to the gradients from other models to the student."], "abstract": "While the success of deep learning relies on large amounts of training datasets, data is often limited in privacy-sensitive domains. To address this challenge, generative model learning with differential privacy has emerged as a solution to train private generative models for desensitized data generation. However, the quality of the images generated by existing methods is limited due to the complexity of modeling data distribution. We build on the success of diffusion models and introduce DP-SAD, which trains a private diffusion model by a stochastic adversarial distillation method. Specifically, we first train a diffusion model as a teacher and then train a student by distillation, in which we achieve differential privacy by adding noise to the gradients from other models to the student. For better generation quality, we introduce a discriminator to distinguish whether an image is from the teacher or the student, which forms the adversarial training. Extensive experiments and analysis clearly demonstrate the effectiveness of our proposed method.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-0d4100aff19b35f4", "title": "Training-Free Activation Sparsity in Large Language Models", "authors": ["James Liu", "Pragaash Ponnusamy", "Tianle Cai", "Han Guo", "Yoon Kim", "Ben Athiwaratkun"], "year": 2024, "published": "2024-08-26T23:30:15Z", "updated": "2025-02-25T21:00:50Z", "arxiv_id": "http://arxiv.org/abs/2408.14690v3", "url": "https://arxiv.org/abs/2408.14690v3", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Memory / RAG"], "benchmarks": [], "models": ["Llama-2", "Llama-3", "Mistral families", "ying"], "key_results": ["TEAL achieves 40-50% model-wide sparsity with minimal performance degradation across Llama-2, Llama-3, and Mistral families, with sizes varying from 7B to 70B.", "We improve existing sparse kernels and demonstrate wall-clock decoding speed-ups of up to 1.53$\\times$ and 1.8$\\times$ at 40% and 50% model-wide sparsity.", "TEAL is compatible with weight quantization, enabling further efficiency gains."], "abstract": "Activation sparsity can enable practical inference speedups in large language models (LLMs) by reducing the compute and memory-movement required for matrix multiplications during the forward pass. However, existing methods face limitations that inhibit widespread adoption. Some approaches are tailored towards older models with ReLU-based sparsity, while others require extensive continued pre-training on up to hundreds of billions of tokens. This paper describes TEAL, a simple training-free method that applies magnitude-based activation sparsity to hidden states throughout the entire model. TEAL achieves 40-50% model-wide sparsity with minimal performance degradation across Llama-2, Llama-3, and Mistral families, with sizes varying from 7B to 70B. We improve existing sparse kernels and demonstrate wall-clock decoding speed-ups of up to 1.53$\\times$ and 1.8$\\times$ at 40% and 50% model-wide sparsity. TEAL is compatible with weight quantization, enabling further efficiency gains.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-1ff31299fb9c865e", "title": "Bridging the Gap: Unpacking the Hidden Challenges in Knowledge Distillation for Online Ranking Systems", "authors": ["Nikhil Khani", "Shuo Yang", "Aniruddh Nath", "Yang Liu", "Pendo Abbo", "Li Wei", "Shawn Andrews", "Maciej Kula", "Jarrod Kahn", "Zhe Zhao", "Lichan Hong", "Ed Chi"], "year": 2024, "published": "2024-08-26T23:01:48Z", "updated": "2024-08-26T23:01:48Z", "arxiv_id": "http://arxiv.org/abs/2408.14678v1", "url": "https://arxiv.org/abs/2408.14678v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.AI", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["ying"], "key_results": ["Our live experiment results demonstrate significant improvements in student model performance while ensuring consistent and reliable generation of high quality teacher labels from a continuous data stream of data."], "abstract": "Knowledge Distillation (KD) is a powerful approach for compressing a large model into a smaller, more efficient model, particularly beneficial for latency-sensitive applications like recommender systems. However, current KD research predominantly focuses on Computer Vision (CV) and NLP tasks, overlooking unique data characteristics and challenges inherent to recommender systems. This paper addresses these overlooked challenges, specifically: (1) mitigating data distribution shifts between teacher and student models, (2) efficiently identifying optimal teacher configurations within time and budgetary constraints, and (3) enabling computationally efficient and rapid sharing of teacher labels to support multiple students. We present a robust KD system developed and rigorously evaluated on multiple large-scale personalized video recommendation systems within Google. Our live experiment results demonstrate significant improvements in student model performance while ensuring consistent and reliable generation of high quality teacher labels from a continuous data stream of data.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-f967258204baabcd", "title": "Revisiting Image Captioning Training Paradigm via Direct CLIP-based Optimization", "authors": ["Nicholas Moratelli", "Davide Caffagni", "Marcella Cornia", "Lorenzo Baraldi", "Rita Cucchiara"], "year": 2024, "published": "2024-08-26T18:00:33Z", "updated": "2024-08-26T18:00:33Z", "arxiv_id": "http://arxiv.org/abs/2408.14547v1", "url": "https://arxiv.org/abs/2408.14547v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.CL", "cs.MM"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "Distillation"], "benchmarks": [], "models": [], "key_results": ["DiCO not only exhibits improved stability and enhanced quality in the generated captions but also aligns more closely with human preferences compared to existing methods, especially in modern metrics."], "abstract": "The conventional training approach for image captioning involves pre-training a network using teacher forcing and subsequent fine-tuning with Self-Critical Sequence Training to maximize hand-crafted captioning metrics. However, when attempting to optimize modern and higher-quality metrics like CLIP-Score and PAC-Score, this training method often encounters instability and fails to acquire the genuine descriptive capabilities needed to produce fluent and informative captions. In this paper, we propose a new training paradigm termed Direct CLIP-Based Optimization (DiCO). Our approach jointly learns and optimizes a reward model that is distilled from a learnable captioning evaluator with high human correlation. This is done by solving a weighted classification problem directly inside the captioner. At the same time, DiCO prevents divergence from the original model, ensuring that fluency is maintained. DiCO not only exhibits improved stability and enhanced quality in the generated captions but also aligns more closely with human preferences compared to existing methods, especially in modern metrics. Additionally, it maintains competitive performance in traditional metrics. Our source code and trained models are publicly available at https://github.com/aimagelab/DiCO.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-3e1f44bff5dc3f2e", "title": "Adaptive Resolution Inference (ARI): Energy-Efficient Machine Learning for Internet of Things", "authors": ["Ziheng Wang", "Pedro Reviriego", "Farzad Niknia", "Javier Conde", "Shanshan Liu", "Fabrizio Lombardi"], "year": 2024, "published": "2024-08-26T16:00:26Z", "updated": "2024-08-26T16:00:26Z", "arxiv_id": "http://arxiv.org/abs/2408.14528v1", "url": "https://arxiv.org/abs/2408.14528v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.DC"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["In recent years, significant efforts have been made to implement simplified ML models that can achieve reasonable performance while reducing computation and energy, for example by pruning weights in neural networks, or using reduced precision for the parameters and arithmetic operations."], "abstract": "The implementation of machine learning in Internet of Things devices poses significant operational challenges due to limited energy and computation resources. In recent years, significant efforts have been made to implement simplified ML models that can achieve reasonable performance while reducing computation and energy, for example by pruning weights in neural networks, or using reduced precision for the parameters and arithmetic operations. However, this type of approach is limited by the performance of the ML implementation, i.e., by the loss for example in accuracy due to the model simplification. In this article, we present adaptive resolution inference (ARI), a novel approach that enables to evaluate new tradeoffs between energy dissipation and model performance in ML implementations. The main principle of the proposed approach is to run inferences with reduced precision (quantization) and use the margin over the decision threshold to determine if either the result is reliable, or the inference must run with the full model. The rationale is that quantization only introduces small deviations in the inference scores, such that if the scores have a sufficient margin over the decision threshold, it is unlikely that the full model would have a different result. Therefore, we can run the quantized model first, and only when the scores do not have a sufficient margin, the full model is run. This enables most inferences to run with the reduced precision model and only a small fraction requires the full model, so significantly reducing computation and energy while not affecting model performance. The proposed ARI approach is presented, analyzed in detail, and evaluated using different data sets for floating-point and stochastic computing implementations. The results show that ARI can significantly reduce the energy for inference in different configurations with savings between 40% and 85%.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-32d136547b3d64b5", "title": "Foundation Models for Music: A Survey", "authors": ["Yinghao Ma", "Anders Øland", "Anton Ragni", "Bleiz MacSen Del Sette", "Charalampos Saitis", "Chris Donahue", "Chenghua Lin", "Christos Plachouras", "Emmanouil Benetos", "Elona Shatri", "Fabio Morreale", "Ge Zhang", "György Fazekas", "Gus Xia", "Huan Zhang", "Ilaria Manco", "Jiawen Huang", "Julien Guinot", "Liwei Lin", "Luca Marinelli", "Max W. Y. Lam", "Megha Sharma", "Qiuqiang Kong", "Roger B. Dannenberg", "Ruibin Yuan", "Shangda Wu", "Shih-Lun Wu", "Shuqi Dai", "Shun Lei", "Shiyin Kang", "Simon Dixon", "Wenhu Chen", "Wenhao Huang", "Xingjian Du", "Xingwei Qu", "Xu Tan", "Yizhi Li", "Zeyue Tian", "Zhiyong Wu", "Zhizheng Wu", "Ziyang Ma", "Ziyu Wang"], "year": 2024, "published": "2024-08-26T15:13:14Z", "updated": "2024-09-03T14:53:34Z", "arxiv_id": "http://arxiv.org/abs/2408.14340v3", "url": "https://arxiv.org/abs/2408.14340v3", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.AI", "cs.CL", "cs.LG", "eess.AS"], "pillar": "Efficient Training & Inference", "methods": ["Interpretability", "Video Generation", "VLM"], "benchmarks": [], "models": [], "key_results": ["This comprehensive review examines state-of-the-art (SOTA) pre-trained models and foundation models in music, spanning from representation learning, generative learning and multimodal learning."], "abstract": "In recent years, foundation models (FMs) such as large language models (LLMs) and latent diffusion models (LDMs) have profoundly impacted diverse sectors, including music. This comprehensive review examines state-of-the-art (SOTA) pre-trained models and foundation models in music, spanning from representation learning, generative learning and multimodal learning. We first contextualise the significance of music in various industries and trace the evolution of AI in music. By delineating the modalities targeted by foundation models, we discover many of the music representations are underexplored in FM development. Then, emphasis is placed on the lack of versatility of previous methods on diverse music applications, along with the potential of FMs in music understanding, generation and medical application. By comprehensively exploring the details of the model pre-training paradigm, architectural choices, tokenisation, finetuning methodologies and controllability, we emphasise the important topics that should have been well explored, like instruction tuning and in-context learning, scaling law and emergent ability, as well as long-sequence modelling etc. A dedicated section presents insights into music agents, accompanied by a thorough analysis of datasets and evaluations essential for pre-training and downstream tasks. Finally, by underscoring the vital importance of ethical considerations, we advocate that following research on FM for music should focus more on such issues as interpretability, transparency, human responsibility, and copyright issues. The paper offers insights into future challenges and trends on FMs for music, aiming to shape the trajectory of human-AI collaboration in the music realm.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-e27d54d92f730c33", "title": "1-Bit FQT: Pushing the Limit of Fully Quantized Training to 1-bit", "authors": ["Chang Gao", "Jianfei Chen", "Kang Zhao", "Jiaqi Wang", "Liping Jing"], "year": 2024, "published": "2024-08-26T13:42:43Z", "updated": "2024-08-26T13:42:43Z", "arxiv_id": "http://arxiv.org/abs/2408.14267v1", "url": "https://arxiv.org/abs/2408.14267v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["To explore the ultimate limit of FQT (the lowest achievable precision), we make a first attempt to 1-bit FQT.", "For fine-tuning VGGNet-16 and ResNet-18 on multiple datasets, our algorithm achieves an average accuracy improvement of approximately 6%, compared to per-sample quantization."], "abstract": "Fully quantized training (FQT) accelerates the training of deep neural networks by quantizing the activations, weights, and gradients into lower precision. To explore the ultimate limit of FQT (the lowest achievable precision), we make a first attempt to 1-bit FQT. We provide a theoretical analysis of FQT based on Adam and SGD, revealing that the gradient variance influences the convergence of FQT. Building on these theoretical results, we introduce an Activation Gradient Pruning (AGP) strategy. The strategy leverages the heterogeneity of gradients by pruning less informative gradients and enhancing the numerical precision of remaining gradients to mitigate gradient variance. Additionally, we propose Sample Channel joint Quantization (SCQ), which utilizes different quantization strategies in the computation of weight gradients and activation gradients to ensure that the method is friendly to low-bitwidth hardware. Finally, we present a framework to deploy our algorithm. For fine-tuning VGGNet-16 and ResNet-18 on multiple datasets, our algorithm achieves an average accuracy improvement of approximately 6%, compared to per-sample quantization. Moreover, our training speedup can reach a maximum of 5.13x compared to full precision training.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-626088308d71a69d", "title": "Scalable Multivariate Fronthaul Quantization for Cell-Free Massive MIMO", "authors": ["Sangwoo Park", "Ahmet Hasim Gokceoglu", "Li Wang", "Osvaldo Simeone"], "year": 2024, "published": "2024-08-26T12:56:41Z", "updated": "2024-08-26T12:56:41Z", "arxiv_id": "http://arxiv.org/abs/2409.06715v1", "url": "https://arxiv.org/abs/2409.06715v1", "primary_category": "cs.IT", "categories": ["cs.IT", "cs.LG", "cs.NI", "eess.SP"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["Previous theoretical work has shown that CP can be potentially improved by a significant margin by precode-and-compress (PC) methods, in which all baseband processing is carried out at the DU, which compresses the precoded signals for transmission on the fronthaul links.", "Numerical results demonstrate that the proposed scalable MQ strategies outperform CP for both the low and high-fronthaul capacity regimes at the cost of increased computational complexity at the DU (but not at the RUs)."], "abstract": "The conventional approach to the fronthaul design for cell-free massive MIMO system follows the compress-and-precode (CP) paradigm. Accordingly, encoded bits and precoding coefficients are shared by the distributed unit (DU) on the fronthaul links, and precoding takes place at the radio units (RUs). Previous theoretical work has shown that CP can be potentially improved by a significant margin by precode-and-compress (PC) methods, in which all baseband processing is carried out at the DU, which compresses the precoded signals for transmission on the fronthaul links. The theoretical performance gain of PC methods are particularly pronounced when the DU implements multivariate quantization (MQ), applying joint quantization across the signals for all the RUs. However, existing solutions for MQ are characterized by a computational complexity that grows exponentially with the sum-fronthaul capacity from the DU to all RUs. This work sets out to design scalable MQ strategies for PC-based cell-free massive MIMO systems. For the low-fronthaul capacity regime, we present alpha-parallel MQ (alpha-PMQ), whose complexity is exponential only in the fronthaul capacity towards an individual RU, while performing close to full MQ. alpha-PMQ tailors MQ to the topology of the network by allowing for parallel local quantization steps for RUs that do not interfere too much with each other. For the high-fronthaul capacity regime, we then introduce neural MQ, which replaces the exhaustive search in MQ with gradient-based updates for a neural-network-based decoder, attaining a complexity that grows linearly with the sum-fronthaul capacity. Numerical results demonstrate that the proposed scalable MQ strategies outperform CP for both the low and high-fronthaul capacity regimes at the cost of increased computational complexity at the DU (but not at the RUs).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7d39ccba83edd075", "title": "TSAK: Two-Stage Semantic-Aware Knowledge Distillation for Efficient Wearable Modality and Model Optimization in Manufacturing Lines", "authors": ["Hymalai Bello", "Daniel Geißler", "Sungho Suh", "Bo Zhou", "Paul Lukowicz"], "year": 2024, "published": "2024-08-26T09:44:21Z", "updated": "2024-08-26T09:44:21Z", "arxiv_id": "http://arxiv.org/abs/2408.14146v1", "url": "https://arxiv.org/abs/2408.14146v1", "primary_category": "cs.LG", "categories": ["cs.LG", "eess.SP"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["We present a two-stage semantic-aware knowledge distillation (KD) approach, TSAK, for efficient, privacy-aware, and wearable HAR in manufacturing lines, which reduces the input sensor modalities as well as the machine learning model size, while reaching similar recognition performance as a larger multi-modal and multi-positional teacher model."], "abstract": "Smaller machine learning models, with less complex architectures and sensor inputs, can benefit wearable sensor-based human activity recognition (HAR) systems in many ways, from complexity and cost to battery life. In the specific case of smart factories, optimizing human-robot collaboration hinges on the implementation of cutting-edge, human-centric AI systems. To this end, workers' activity recognition enables accurate quantification of performance metrics, improving efficiency holistically. We present a two-stage semantic-aware knowledge distillation (KD) approach, TSAK, for efficient, privacy-aware, and wearable HAR in manufacturing lines, which reduces the input sensor modalities as well as the machine learning model size, while reaching similar recognition performance as a larger multi-modal and multi-positional teacher model. The first stage incorporates a teacher classifier model encoding attention, causal, and combined representations. The second stage encompasses a semantic classifier merging the three representations from the first stage. To evaluate TSAK, we recorded a multi-modal dataset at a smart factory testbed with wearable and privacy-aware sensors (IMU and capacitive) located on both workers' hands. In addition, we evaluated our approach on OpenPack, the only available open dataset mimicking the wearable sensor placements on both hands in the manufacturing HAR scenario. We compared several KD strategies with different representations to regulate the training process of a smaller student model. Compared to the larger teacher model, the student model takes fewer sensor channels from a single hand, has 79% fewer parameters, runs 8.88 times faster, and requires 96.6% less computing power (FLOPS).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7800a234a6531eb4", "title": "Exploring the Potential of Large Language Models for Heterophilic Graphs", "authors": ["Yuxia Wu", "Shujie Li", "Yuan Fang", "Chuan Shi"], "year": 2024, "published": "2024-08-26T09:29:56Z", "updated": "2025-02-15T12:53:29Z", "arxiv_id": "http://arxiv.org/abs/2408.14134v3", "url": "https://arxiv.org/abs/2408.14134v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL", "cs.SI"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": ["philic", "ying"], "key_results": ["Large language models (LLMs) have presented significant opportunities to enhance various machine learning applications, including graph neural networks (GNNs)."], "abstract": "Large language models (LLMs) have presented significant opportunities to enhance various machine learning applications, including graph neural networks (GNNs). By leveraging the vast open-world knowledge within LLMs, we can more effectively interpret and utilize textual data to better characterize heterophilic graphs, where neighboring nodes often have different labels. However, existing approaches for heterophilic graphs overlook the rich textual data associated with nodes, which could unlock deeper insights into their heterophilic contexts. In this work, we explore the potential of LLMs for modeling heterophilic graphs and propose a novel two-stage framework: LLM-enhanced edge discriminator and LLM-guided edge reweighting. In the first stage, we fine-tune the LLM to better identify homophilic and heterophilic edges based on the textual content of their nodes. In the second stage, we adaptively manage message propagation in GNNs for different edge types based on node features, structures, and heterophilic or homophilic characteristics. To cope with the computational demands when deploying LLMs in practical scenarios, we further explore model distillation techniques to fine-tune smaller, more efficient models that maintain competitive performance. Extensive experiments validate the effectiveness of our framework, demonstrating the feasibility of using LLMs to enhance node classification on heterophilic graphs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-cc398318dd83e208", "title": "GlaLSTM: A Concurrent LSTM Stream Framework for Glaucoma Detection via Biomarker Mining", "authors": ["Cheng Huang", "Weizheng Xie", "Tsengdar Lee", "Karanjit Kooner", "Ning Zhang", "Jia Zhang"], "year": 2024, "published": "2024-08-28T06:08:46Z", "updated": "2025-09-22T19:09:54Z", "arxiv_id": "http://arxiv.org/abs/2408.15555v3", "url": "https://arxiv.org/abs/2408.15555v3", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["This approach not only improves detection accuracy but also empowers clinicians with actionable insights, facilitating more informed decision-making.", "Experimental evaluations confirm that GlaLSTM surpasses existing state-of-the-art methods, demonstrating its potential for both advanced biomarker analysis and reliable glaucoma detection."], "abstract": "Glaucoma is a complex group of eye diseases marked by optic nerve damage, commonly linked to elevated intraocular pressure and biomarkers like retinal nerve fiber layer thickness. Understanding how these biomarkers interact is crucial for unraveling glaucoma's underlying mechanisms. In this paper, we propose GlaLSTM, a novel concurrent LSTM stream framework for glaucoma detection, leveraging latent biomarker relationships. Unlike traditional CNN-based models that primarily detect glaucoma from images, GlaLSTM provides deeper interpretability, revealing the key contributing factors and enhancing model transparency. This approach not only improves detection accuracy but also empowers clinicians with actionable insights, facilitating more informed decision-making. Experimental evaluations confirm that GlaLSTM surpasses existing state-of-the-art methods, demonstrating its potential for both advanced biomarker analysis and reliable glaucoma detection.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-db39dc87f975ad98", "title": "WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback", "authors": ["Taiwei Shi", "Zhuoer Wang", "Longqi Yang", "Ying-Chun Lin", "Zexue He", "Mengting Wan", "Pei Zhou", "Sujay Jauhar", "Sihao Chen", "Shan Xia", "Hongfei Zhang", "Jieyu Zhao", "Xiaofeng Xu", "Xia Song", "Jennifer Neville"], "year": 2024, "published": "2024-08-28T05:53:46Z", "updated": "2026-04-17T16:47:12Z", "arxiv_id": "http://arxiv.org/abs/2408.15549v4", "url": "https://arxiv.org/abs/2408.15549v4", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["Our experiments demonstrate that LLMs fine-tuned on WildFeedback dataset exhibit significantly improved alignment with user preferences, as evidenced by both traditional benchmarks and our proposed checklist-guided evaluation."], "abstract": "As large language models (LLMs) continue to advance, aligning these models with human preferences has emerged as a critical challenge. Traditional alignment methods, relying on human or LLM annotated datasets, are limited by their resource-intensive nature, inherent subjectivity, misalignment with real-world user preferences, and the risk of feedback loops that amplify model biases. To overcome these limitations, we introduce WildFeedback, a novel framework that leverages in-situ user feedback during conversations with LLMs to create preference datasets automatically. Given a corpus of multi-turn user-LLM conversation, WildFeedback identifies and classifies user feedback to LLM responses between conversation turns. The user feedback is then used to create examples of preferred and dispreferred responses according to users' preference. Our experiments demonstrate that LLMs fine-tuned on WildFeedback dataset exhibit significantly improved alignment with user preferences, as evidenced by both traditional benchmarks and our proposed checklist-guided evaluation. By incorporating in-situ feedback from actual users, WildFeedback addresses the scalability, subjectivity, and bias challenges that plague existing approaches, marking a significant step toward developing LLMs that are more responsive to the diverse and evolving needs of their users.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-1be8cc59b628dfd3", "title": "PersonalizedUS: Interpretable Breast Cancer Risk Assessment with Local Coverage Uncertainty Quantification", "authors": ["Alek Fröhlich", "Thiago Ramos", "Gustavo Cabello", "Isabela Buzatto", "Rafael Izbicki", "Daniel Tiezzi"], "year": 2024, "published": "2024-08-28T00:47:55Z", "updated": "2024-08-28T00:47:55Z", "arxiv_id": "http://arxiv.org/abs/2408.15458v1", "url": "https://arxiv.org/abs/2408.15458v1", "primary_category": "cs.LG", "categories": ["cs.LG", "eess.IV", "stat.ML"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["Moreover, we make available a curated tabular dataset of 1936 biopsied breast lesions from a recent observational multicenter study and benchmark the performance of several state-of-the-art learning algorithms."], "abstract": "Correctly assessing the malignancy of breast lesions identified during ultrasound examinations is crucial for effective clinical decision-making. However, the current \"golden standard\" relies on manual BI-RADS scoring by clinicians, often leading to unnecessary biopsies and a significant mental health burden on patients and their families. In this paper, we introduce PersonalizedUS, an interpretable machine learning system that leverages recent advances in conformal prediction to provide precise and personalized risk estimates with local coverage guarantees and sensitivity, specificity, and predictive values above 0.9 across various threshold levels. In particular, we identify meaningful lesion subgroups where distribution-free, model-agnostic conditional coverage holds, with approximately 90% of our prediction sets containing only the ground truth in most lesion subgroups, thus explicitly characterizing for which patients the model is most suitably applied. Moreover, we make available a curated tabular dataset of 1936 biopsied breast lesions from a recent observational multicenter study and benchmark the performance of several state-of-the-art learning algorithms. We also report a successful case study of the deployed system in the same multicenter context. Concrete clinical benefits include up to a 65% reduction in requested biopsies among BI-RADS 4a and 4b lesions, with minimal to no missed cancer cases.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-74151a62d69e324f", "title": "Understanding GNNs for Boolean Satisfiability through Approximation Algorithms", "authors": ["Jan Hůla", "David Mojžíšek", "Mikoláš Janota"], "year": 2024, "published": "2024-08-27T21:47:09Z", "updated": "2024-08-27T21:47:09Z", "arxiv_id": "http://arxiv.org/abs/2408.15418v1", "url": "https://arxiv.org/abs/2408.15418v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["We show that the curriculum, together with several other optimizations, reduces the training time by more than an order of magnitude compared to the baseline without the curriculum."], "abstract": "The paper deals with the interpretability of Graph Neural Networks in the context of Boolean Satisfiability. The goal is to demystify the internal workings of these models and provide insightful perspectives into their decision-making processes. This is done by uncovering connections to two approximation algorithms studied in the domain of Boolean Satisfiability: Belief Propagation and Semidefinite Programming Relaxations. Revealing these connections has empowered us to introduce a suite of impactful enhancements. The first significant enhancement is a curriculum training procedure, which incrementally increases the problem complexity in the training set, together with increasing the number of message passing iterations of the Graph Neural Network. We show that the curriculum, together with several other optimizations, reduces the training time by more than an order of magnitude compared to the baseline without the curriculum. Furthermore, we apply decimation and sampling of initial embeddings, which significantly increase the percentage of solved problems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-0cc63e52a529cb48", "title": "Conformal Disentanglement and Latent-Space Curation: A Neural Framework for Perspective Synthesis, Differentiation and Targeted Generation", "authors": ["George A. Kevrekidis", "Eleni D. Koronaki", "Dimitris G. Giovanis", "Yannis G. Kevrekidis"], "year": 2024, "published": "2024-08-27T18:06:45Z", "updated": "2026-06-04T22:55:13Z", "arxiv_id": "http://arxiv.org/abs/2408.15344v2", "url": "https://arxiv.org/abs/2408.15344v2", "primary_category": "cs.LG", "categories": ["cs.LG", "math.DS"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": ["MATH"], "models": ["ying", "yielding"], "key_results": ["Many scientific and engineering problems involve observing a common phenomenon through multiple heterogeneous sensors or measurement modalities."], "abstract": "Many scientific and engineering problems involve observing a common phenomenon through multiple heterogeneous sensors or measurement modalities. Such observations typically contain both information shared across sensors, reflecting the underlying system, and sensor-specific or extraneous components arising from measurement processes or environmental effects. Disentangling these contributions is essential when sensor-independent observations are unavailable. We propose a neural autoencoder framework that explicitly separates shared and sensor-specific latent variables from multi-sensor data. The architecture enforces geometric independence between latent components through structural constraints and orthogonality-based regularization, yielding interpretable and disentangled representations. Building on this representation, we then introduce a latent-space generative methodology in which generative models are tuned/\"restricted\" on selected disentangled latent subspaces; we then constructively combine disentangled observed latent variables to conditionally synthesize new samples via trained decoders. This enables consistent data generation with prescribed shared (or sensor-specific) characteristics. It also supports cross-sensor inference by consistently sampling distributions over plausible measurements in unobserved modalities. We demonstrate the approach on several computational examples, showing effective disentanglement, targeted data generation, and modality imputation in heterogeneous sensing settings.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7a6c033b409d3800", "title": "UNA: A Unified Supervised Framework for Efficient LLM Alignment Across Feedback Types", "authors": ["Zhichao Wang", "Bin Bi", "Can Huang", "Shiva Kumar Pentyala", "Zixu James Zhu", "Sitaram Asur", "Na Claire Cheng", "Cheng Wan", "Dong Nie", "Lingzi Hong"], "year": 2024, "published": "2024-08-27T18:04:07Z", "updated": "2026-05-07T20:32:48Z", "arxiv_id": "http://arxiv.org/abs/2408.15339v4", "url": "https://arxiv.org/abs/2408.15339v4", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["RL alignment methods, including RLHF and DPO, are primarily based on pairwise preference data."], "abstract": "RL alignment methods, including RLHF and DPO, are primarily based on pairwise preference data. Although scalar or score-based feedback has been collected in some settings, it is rarely used directly, and preference magnitude information is typically ignored. Furthermore, current alignment frameworks offer limited capability for unifying heterogeneous supervision signals, making it difficult to jointly leverage diverse data types within a single training paradigm. This limitation constrains the richness and scalability of the alignment process. To address this gap, we propose a \\textbf{UN}ified \\textbf{A}lignment (UNA) framework capable of training across different types of feedback, including binary, pairwise, and score-based, through a generalized implicit reward function. The reward function is theoretically proved to be the optimal policy by the log sum inequality. Extensive experiments on classical benchmarks consistently demonstrate the advantage of the proposed unified framework with typical LLM base models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-382e6f1e6a9bf615", "title": "LLM Defenses Are Not Robust to Multi-Turn Human Jailbreaks Yet", "authors": ["Nathaniel Li", "Ziwen Han", "Ian Steneker", "Willow Primack", "Riley Goodside", "Hugh Zhang", "Zifan Wang", "Cristina Menghini", "Summer Yue"], "year": 2024, "published": "2024-08-27T17:33:30Z", "updated": "2024-09-04T00:58:59Z", "arxiv_id": "http://arxiv.org/abs/2408.15221v2", "url": "https://arxiv.org/abs/2408.15221v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL", "cs.CR", "cs.CY"], "pillar": "RL / Alignment / Safety", "methods": ["Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Recent large language model (LLM) defenses have greatly improved models' ability to refuse harmful queries, even when adversarially attacked.", "However, LLM defenses are primarily evaluated against automated adversarial attacks in a single turn of conversation, an insufficient threat model for real-world malicious use.", "We demonstrate that multi-turn human jailbreaks uncover significant vulnerabilities, exceeding 70% attack success rate (ASR) on HarmBench against defenses that report single-digit ASRs with automated single-turn attacks."], "abstract": "Recent large language model (LLM) defenses have greatly improved models' ability to refuse harmful queries, even when adversarially attacked. However, LLM defenses are primarily evaluated against automated adversarial attacks in a single turn of conversation, an insufficient threat model for real-world malicious use. We demonstrate that multi-turn human jailbreaks uncover significant vulnerabilities, exceeding 70% attack success rate (ASR) on HarmBench against defenses that report single-digit ASRs with automated single-turn attacks. Human jailbreaks also reveal vulnerabilities in machine unlearning defenses, successfully recovering dual-use biosecurity knowledge from unlearned models. We compile these results into Multi-Turn Human Jailbreaks (MHJ), a dataset of 2,912 prompts across 537 multi-turn jailbreaks. We publicly release MHJ alongside a compendium of jailbreak tactics developed across dozens of commercial red teaming engagements, supporting research towards stronger LLM defenses.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-36f4c8e599b978db", "title": "Improving Adversarial Robustness in Android Malware Detection by Reducing the Impact of Spurious Correlations", "authors": ["Hamid Bostani", "Zhengyu Zhao", "Veelasha Moonsamy"], "year": 2024, "published": "2024-08-27T17:01:12Z", "updated": "2024-08-27T17:01:12Z", "arxiv_id": "http://arxiv.org/abs/2408.16025v1", "url": "https://arxiv.org/abs/2408.16025v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG", "cs.SE"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Machine learning (ML) has demonstrated significant advancements in Android malware detection (AMD); however, the resilience of ML against realistic evasion attacks remains a major obstacle for AMD.", "In this study, we propose a domain adaptation technique to improve the generalizability of AMD by aligning the distribution of malware samples and AEs.", "Our experiments conducted on DREBIN, a renowned Android malware detector, demonstrate that our approach surpasses the state-of-the-art defense, Sec-SVM, when facing realistic evasion attacks."], "abstract": "Machine learning (ML) has demonstrated significant advancements in Android malware detection (AMD); however, the resilience of ML against realistic evasion attacks remains a major obstacle for AMD. One of the primary factors contributing to this challenge is the scarcity of reliable generalizations. Malware classifiers with limited generalizability tend to overfit spurious correlations derived from biased features. Consequently, adversarial examples (AEs), generated by evasion attacks, can modify these features to evade detection. In this study, we propose a domain adaptation technique to improve the generalizability of AMD by aligning the distribution of malware samples and AEs. Specifically, we utilize meaningful feature dependencies, reflecting domain constraints in the feature space, to establish a robust feature space. Training on the proposed robust feature space enables malware classifiers to learn from predefined patterns associated with app functionality rather than from individual features. This approach helps mitigate spurious correlations inherent in the initial feature space. Our experiments conducted on DREBIN, a renowned Android malware detector, demonstrate that our approach surpasses the state-of-the-art defense, Sec-SVM, when facing realistic evasion attacks. In particular, our defense can improve adversarial robustness by up to 55% against realistic evasion attacks compared to Sec-SVM.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-ee8a7ad2571ad776", "title": "On latent dynamics learning in nonlinear reduced order modeling", "authors": ["Nicola Farenga", "Stefania Fresca", "Simone Brivio", "Andrea Manzoni"], "year": 2024, "published": "2024-08-27T16:35:06Z", "updated": "2024-11-28T11:53:43Z", "arxiv_id": "http://arxiv.org/abs/2408.15183v2", "url": "https://arxiv.org/abs/2408.15183v2", "primary_category": "math.NA", "categories": ["math.NA", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Video Generation"], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["In this work, we present the novel mathematical framework of latent dynamics models (LDMs) for reduced order modeling of parameterized nonlinear time-dependent PDEs."], "abstract": "In this work, we present the novel mathematical framework of latent dynamics models (LDMs) for reduced order modeling of parameterized nonlinear time-dependent PDEs. Our framework casts this latter task as a nonlinear dimensionality reduction problem, while constraining the latent state to evolve accordingly to an (unknown) dynamical system. A time-continuous setting is employed to derive error and stability estimates for the LDM approximation of the full order model (FOM) solution. We analyze the impact of using an explicit Runge-Kutta scheme in the time-discrete setting, resulting in the $Δ\\text{LDM}$ formulation, and further explore the learnable setting, $Δ\\text{LDM}_θ$, where deep neural networks approximate the discrete LDM components, while providing a bounded approximation error with respect to the FOM. Moreover, we extend the concept of parameterized Neural ODE - recently proposed as a possible way to build data-driven dynamical systems with varying input parameters - to be a convolutional architecture, where the input parameters information is injected by means of an affine modulation mechanism, while designing a convolutional autoencoder neural network able to retain spatial-coherence, thus enhancing interpretability at the latent level. Numerical experiments, including the Burgers' and the advection-reaction-diffusion equations, demonstrate the framework's ability to obtain, in a multi-query context, a time-continuous approximation of the FOM solution, thus being able to query the LDM approximation at any given time instance while retaining a prescribed level of accuracy. Our findings highlight the remarkable potential of the proposed LDMs, representing a mathematically rigorous framework to enhance the accuracy and approximation capabilities of reduced order modeling for time-dependent parameterized PDEs.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4911505bca8c2ffe", "title": "Multi-Modal Instruction-Tuning Small-Scale Language-and-Vision Assistant for Semiconductor Electron Micrograph Analysis", "authors": ["Sakhinana Sagar Srinivas", "Geethan Sannidhi", "Venkataramana Runkana"], "year": 2024, "published": "2024-08-27T15:50:04Z", "updated": "2024-08-27T15:50:04Z", "arxiv_id": "http://arxiv.org/abs/2409.07463v1", "url": "https://arxiv.org/abs/2409.07463v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["VLM"], "benchmarks": [], "models": ["GPT-4"], "key_results": ["We present a novel framework for analyzing and interpreting electron microscopy images in semiconductor manufacturing using vision-language instruction tuning."], "abstract": "We present a novel framework for analyzing and interpreting electron microscopy images in semiconductor manufacturing using vision-language instruction tuning. The framework employs a unique teacher-student approach, leveraging pre-trained multimodal large language models such as GPT-4 to generate instruction-following data for zero-shot visual question answering (VQA) and classification tasks, customizing smaller multimodal models (SMMs) for microscopy image analysis, resulting in an instruction-tuned language-and-vision assistant. Our framework merges knowledge engineering with machine learning to integrate domain-specific expertise from larger to smaller multimodal models within this specialized field, greatly reducing the need for extensive human labeling. Our study presents a secure, cost-effective, and customizable approach for analyzing microscopy images, addressing the challenges of adopting proprietary models in semiconductor manufacturing.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-6331b85c967a56c6", "title": "How transformers learn structured data: insights from hierarchical filtering", "authors": ["Jerome Garnier-Brun", "Marc Mézard", "Emanuele Moscato", "Luca Saglietti"], "year": 2024, "published": "2024-08-27T15:23:09Z", "updated": "2025-06-10T08:51:44Z", "arxiv_id": "http://arxiv.org/abs/2408.15138v3", "url": "https://arxiv.org/abs/2408.15138v3", "primary_category": "cs.LG", "categories": ["cs.LG", "cond-mat.dis-nn", "cond-mat.stat-mech", "cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["Understanding the learning process and the embedded computation in transformers is becoming a central goal for the development of interpretable AI."], "abstract": "Understanding the learning process and the embedded computation in transformers is becoming a central goal for the development of interpretable AI. In the present study, we introduce a hierarchical filtering procedure for data models of sequences on trees, allowing us to hand-tune the range of positional correlations in the data. Leveraging this controlled setting, we provide evidence that vanilla encoder-only transformers can approximate the exact inference algorithm when trained on root classification and masked language modeling tasks, and study how this computation is discovered and implemented. We find that correlations at larger distances, corresponding to increasing layers of the hierarchy, are sequentially included by the network during training. By comparing attention maps from models trained with varying degrees of filtering and by probing the different encoder levels, we find clear evidence of a reconstruction of correlations on successive length scales corresponding to the various levels of the hierarchy, which we relate to a plausible implementation of the exact inference algorithm within the same architecture.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-771d97de58ef377d", "title": "TagOOD: A Novel Approach to Out-of-Distribution Detection via Vision-Language Representations and Class Center Learning", "authors": ["Jinglun Li", "Xinyu Zhou", "Kaixun Jiang", "Lingyi Hong", "Pinxue Guo", "Zhaoyu Chen", "Weifeng Ge", "Wenqiang Zhang"], "year": 2024, "published": "2024-08-28T06:37:59Z", "updated": "2024-08-28T06:37:59Z", "arxiv_id": "http://arxiv.org/abs/2408.15566v1", "url": "https://arxiv.org/abs/2408.15566v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["This enriched data representation improves performance across various tasks.", "In this paper, we propose \\textbf{TagOOD}, a novel approach for OOD detection that leverages vision-language representations to achieve label-free object feature decoupling from whole images."], "abstract": "Multimodal fusion, leveraging data like vision and language, is rapidly gaining traction. This enriched data representation improves performance across various tasks. Existing methods for out-of-distribution (OOD) detection, a critical area where AI models encounter unseen data in real-world scenarios, rely heavily on whole-image features. These image-level features can include irrelevant information that hinders the detection of OOD samples, ultimately limiting overall performance. In this paper, we propose \\textbf{TagOOD}, a novel approach for OOD detection that leverages vision-language representations to achieve label-free object feature decoupling from whole images. This decomposition enables a more focused analysis of object semantics, enhancing OOD detection performance. Subsequently, TagOOD trains a lightweight network on the extracted object features to learn representative class centers. These centers capture the central tendencies of IND object classes, minimizing the influence of irrelevant image features during OOD detection. Finally, our approach efficiently detects OOD samples by calculating distance-based metrics as OOD scores between learned centers and test samples. We conduct extensive experiments to evaluate TagOOD on several benchmark datasets and demonstrate its superior performance compared to existing OOD detection methods. This work presents a novel perspective for further exploration of multimodal information utilization in OOD detection, with potential applications across various tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-8397edb44c8d8edf", "title": "Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models", "authors": ["Wenbin Wang", "Liang Ding", "Minyan Zeng", "Xiabin Zhou", "Li Shen", "Yong Luo", "Dacheng Tao"], "year": 2024, "published": "2024-08-28T06:09:02Z", "updated": "2024-08-28T06:09:02Z", "arxiv_id": "http://arxiv.org/abs/2408.15556v1", "url": "https://arxiv.org/abs/2408.15556v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["While state-of-the-art (SOTA) MLLMs claim to process images at 4K resolution, existing MLLM benchmarks only support up to 2K, leaving the capabilities of SOTA models on true HR images largely untested.", "Extensive experiments show that: 1) the SOTA MLLM achieves 63% accuracy, which is markedly lower than the 87% accuracy achieved by humans on HR-Bench; 2) our DC$^2$ brings consistent and significant improvements (a relative increase of +6% on HR-Bench and +8% on general multimodal benchmarks)."], "abstract": "Multimodal large language models (MLLMs) have experienced significant advancements recently, but still struggle to recognize and interpret intricate details in high-resolution (HR) images effectively. While state-of-the-art (SOTA) MLLMs claim to process images at 4K resolution, existing MLLM benchmarks only support up to 2K, leaving the capabilities of SOTA models on true HR images largely untested. Furthermore, existing methods for enhancing HR image perception in MLLMs rely on computationally expensive visual instruction tuning. To address these limitations, we introduce HR-Bench, the first deliberately designed benchmark to rigorously evaluate MLLM performance on 4K&8K images. Through extensive experiments, we demonstrate that while downsampling HR images leads to vision information loss, leveraging complementary modalities, e.g., text, can effectively compensate for this loss. Building upon this insight, we propose Divide, Conquer and Combine (DC$^2$), a novel training-free framework for enhancing MLLM perception of HR images. DC$^2$ follows a three-staged approach: 1) Divide: recursively partitioning the HR image into patches and merging similar patches to minimize computational overhead, 2) Conquer: leveraging the MLLM to generate accurate textual descriptions for each image patch, and 3) Combine: utilizing the generated text descriptions to enhance the MLLM's understanding of the overall HR image. Extensive experiments show that: 1) the SOTA MLLM achieves 63% accuracy, which is markedly lower than the 87% accuracy achieved by humans on HR-Bench; 2) our DC$^2$ brings consistent and significant improvements (a relative increase of +6% on HR-Bench and +8% on general multimodal benchmarks). The benchmark and code will be released to facilitate the multimodal R&D community.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-22c2e6009beab986", "title": "Kangaroo: A Powerful Video-Language Model Supporting Long-context Video Input", "authors": ["Jiajun Liu", "Yibing Wang", "Hanghang Ma", "Xiaoping Wu", "Xiaoqi Ma", "Xiaoming Wei", "Jianbin Jiao", "Enhua Wu", "Jie Hu"], "year": 2024, "published": "2024-08-28T05:34:14Z", "updated": "2024-08-28T05:34:14Z", "arxiv_id": "http://arxiv.org/abs/2408.15542v1", "url": "https://arxiv.org/abs/2408.15542v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["Long Context", "VLM"], "benchmarks": [], "models": [], "key_results": ["Evaluation results demonstrate that, with 8B parameters, Kangaroo achieves state-of-the-art performance across a variety of video understanding benchmarks while exhibiting competitive results on others."], "abstract": "Rapid advancements have been made in extending Large Language Models (LLMs) to Large Multi-modal Models (LMMs). However, extending input modality of LLMs to video data remains a challenging endeavor, especially for long videos. Due to insufficient access to large-scale high-quality video data and the excessive compression of visual features, current methods exhibit limitations in effectively processing long videos. In this paper, we introduce Kangaroo, a powerful Video LMM aimed at addressing these challenges. Confronted with issue of inadequate training data, we develop a data curation system to build a large-scale dataset with high-quality annotations for vision-language pre-training and instruction tuning. In addition, we design a curriculum training pipeline with gradually increasing resolution and number of input frames to accommodate long videos. Evaluation results demonstrate that, with 8B parameters, Kangaroo achieves state-of-the-art performance across a variety of video understanding benchmarks while exhibiting competitive results on others. Particularly, on benchmarks specialized for long videos, Kangaroo excels some larger models with over 10B parameters and proprietary models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-2c3831c6b119e208", "title": "A Simple Baseline with Single-encoder for Referring Image Segmentation", "authors": ["Seonghoon Yu", "Ilchae Jung", "Byeongju Han", "Taeoh Kim", "Yunho Kim", "Dongyoon Wee", "Jeany Son"], "year": 2024, "published": "2024-08-28T04:14:01Z", "updated": "2025-06-17T06:34:15Z", "arxiv_id": "http://arxiv.org/abs/2408.15521v3", "url": "https://arxiv.org/abs/2408.15521v3", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["BERT"], "key_results": ["Our simple baseline with a single encoder achieves outstanding performances on the RIS benchmark datasets while maintaining computational efficiency, compared to the most recent SoTA methods based on dual-encoders."], "abstract": "Referring image segmentation (RIS) requires dense vision-language interactions between visual pixels and textual words to segment objects based on a given description. However, commonly adapted dual-encoders in RIS, e.g., Swin transformer and BERT (uni-modal encoders) or CLIP (a multi-modal dual-encoder), lack dense multi-modal interactions during pre-training, leading to a gap with a pixel-level RIS task. To bridge this gap, existing RIS methods often rely on multi-modal fusion modules that interact two encoders, but this approach leads to high computational costs. In this paper, we present a novel RIS method with a single-encoder, i.e., BEiT-3, maximizing the potential of shared self-attention across all framework components. This enables seamless interactions of two modalities from input to final prediction, producing granularly aligned multi-modal features. Furthermore, we propose lightweight yet effective decoder modules, a Shared FPN and a Shared Mask Decoder, which contribute to the high efficiency of our model. Our simple baseline with a single encoder achieves outstanding performances on the RIS benchmark datasets while maintaining computational efficiency, compared to the most recent SoTA methods based on dual-encoders.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-01a2c5d633d401fe", "title": "Meta-Learn Unimodal Signals with Weak Supervision for Multimodal Sentiment Analysis", "authors": ["Sijie Mai", "Yu Zhao", "Ying Zeng", "Jianhua Yao", "Haifeng Hu"], "year": 2024, "published": "2024-08-28T03:43:01Z", "updated": "2024-09-13T02:51:18Z", "arxiv_id": "http://arxiv.org/abs/2408.16029v2", "url": "https://arxiv.org/abs/2408.16029v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["However, there exists the noisy label problem for the learning of unimodal signals as multimodal annotations are not always the ideal substitutes for the unimodal ones, failing to achieve finer optimization for individual modalities.", "Experimental results suggest that MUG outperforms competitive baselines and can learn accurate unimodal labels."], "abstract": "Multimodal sentiment analysis aims to effectively integrate information from various sources to infer sentiment, where in many cases there are no annotations for unimodal labels. Therefore, most works rely on multimodal labels for training. However, there exists the noisy label problem for the learning of unimodal signals as multimodal annotations are not always the ideal substitutes for the unimodal ones, failing to achieve finer optimization for individual modalities. In this paper, we explore the learning of unimodal labels under the weak supervision from the annotated multimodal labels. Specifically, we propose a novel meta uni-label generation (MUG) framework to address the above problem, which leverages the available multimodal labels to learn the corresponding unimodal labels by the meta uni-label correction network (MUCN). We first design a contrastive-based projection module to bridge the gap between unimodal and multimodal representations, so as to use multimodal annotations to guide the learning of MUCN. Afterwards, we propose unimodal and multimodal denoising tasks to train MUCN with explicit supervision via a bi-level optimization strategy. We then jointly train unimodal and multimodal learning tasks to extract discriminative unimodal features for multimodal inference. Experimental results suggest that MUG outperforms competitive baselines and can learn accurate unimodal labels.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-a17983e47f7a17cf", "title": "Hand1000: Generating Realistic Hands from Text with Only 1,000 Images", "authors": ["Haozhuo Zhang", "Bin Zhu", "Yu Cao", "Yanbin Hao"], "year": 2024, "published": "2024-08-28T00:54:51Z", "updated": "2024-12-21T01:42:38Z", "arxiv_id": "http://arxiv.org/abs/2408.15461v3", "url": "https://arxiv.org/abs/2408.15461v3", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Video Generation"], "benchmarks": [], "models": ["LLaMA3", "Stable Diffusion model to generate realistic hand images. In addition"], "key_results": ["Text-to-image generation models have achieved remarkable advancements in recent years, aiming to produce realistic images from textual descriptions.", "The second stage further optimizes text embedding by incorporating the extracted hand gesture representation, to improve alignment between the textual descriptions and the generated hand images.", "Extensive experiments demonstrate that Hand1000 significantly outperforms existing models in producing anatomically correct hand images while faithfully representing other details in the text, such as faces, clothing, and colors."], "abstract": "Text-to-image generation models have achieved remarkable advancements in recent years, aiming to produce realistic images from textual descriptions. However, these models often struggle with generating anatomically accurate representations of human hands. The resulting images frequently exhibit issues such as incorrect numbers of fingers, unnatural twisting or interlacing of fingers, or blurred and indistinct hands. These issues stem from the inherent complexity of hand structures and the difficulty in aligning textual descriptions with precise visual depictions of hands. To address these challenges, we propose a novel approach named Hand1000 that enables the generation of realistic hand images with target gesture using only 1,000 training samples. The training of Hand1000 is divided into three stages with the first stage aiming to enhance the model's understanding of hand anatomy by using a pre-trained hand gesture recognition model to extract gesture representation. The second stage further optimizes text embedding by incorporating the extracted hand gesture representation, to improve alignment between the textual descriptions and the generated hand images. The third stage utilizes the optimized embedding to fine-tune the Stable Diffusion model to generate realistic hand images. In addition, we construct the first publicly available dataset specifically designed for text-to-hand image generation. Based on the existing hand gesture recognition dataset, we adopt advanced image captioning models and LLaMA3 to generate high-quality textual descriptions enriched with detailed gesture information. Extensive experiments demonstrate that Hand1000 significantly outperforms existing models in producing anatomically correct hand images while faithfully representing other details in the text, such as faces, clothing, and colors.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-09d30607f5517543", "title": "DualKanbaFormer: An Efficient Selective Sparse Framework for Multimodal Aspect-based Sentiment Analysis", "authors": ["Adamu Lawan", "Juhua Pu", "Haruna Yunusa", "Muhammad Lawan", "Aliyu Umar", "Adamu Sani Yahya", "Mahmoud Basi"], "year": 2024, "published": "2024-08-27T19:33:15Z", "updated": "2025-04-19T17:40:47Z", "arxiv_id": "http://arxiv.org/abs/2408.15379v3", "url": "https://arxiv.org/abs/2408.15379v3", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Comprehensive experiments on two publicly available datasets reveal that DualKanbaFormer consistently outperforms several state-of-the-art (SOTA) models."], "abstract": "Multimodal Aspect-based Sentiment Analysis (MABSA) enhances sentiment detection by integrating textual data with complementary modalities, such as images, to provide a more refined and comprehensive understanding of sentiment. However, conventional attention mechanisms, despite notable benchmarks, are hindered by quadratic complexity, limiting their ability to fully capture global contextual dependencies and rich semantic information in both modalities. To address this limitation, we introduce DualKanbaFormer, a novel framework that leverages parallel Textual and Visual KanbaFormer modules for robust multimodal analysis. Our approach incorporates Aspect-Driven Sparse Attention (ADSA) to dynamically balance coarse-grained aggregation and fine-grained selection for aspect-focused precision, ensuring the preservation of both global context awareness and local precision in textual and visual representations. Additionally, we utilize the Selective State Space Model (Mamba) to capture extensive global semantic information across both modalities. Furthermore, We replace traditional feed-forward networks and normalization with Kolmogorov-Arnold Networks (KANs) and Dynamic Tanh (DyT) to enhance non-linear expressivity and inference stability. To facilitate the effective integration of textual and visual features, we design a multimodal gated fusion layer that dynamically optimizes inter-modality interactions, significantly enhancing the models efficacy in MABSA tasks. Comprehensive experiments on two publicly available datasets reveal that DualKanbaFormer consistently outperforms several state-of-the-art (SOTA) models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4e3afdd67025c172", "title": "GenRec: Unifying Video Generation and Recognition with Diffusion Models", "authors": ["Zejia Weng", "Xitong Yang", "Zhen Xing", "Zuxuan Wu", "Yu-Gang Jiang"], "year": 2024, "published": "2024-08-27T17:59:41Z", "updated": "2024-11-12T06:08:29Z", "arxiv_id": "http://arxiv.org/abs/2408.15241v2", "url": "https://arxiv.org/abs/2408.15241v2", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": ["ying"], "key_results": ["In particular, GenRec achieves competitive recognition performance, offering 75.8% and 87.2% accuracy on SSV2 and K400, respectively.", "GenRec also performs the best on class-conditioned image-to-video generation, achieving 46.5 and 49.3 FVD scores on SSV2 and EK-100 datasets."], "abstract": "Video diffusion models are able to generate high-quality videos by learning strong spatial-temporal priors on large-scale datasets. In this paper, we aim to investigate whether such priors derived from a generative process are suitable for video recognition, and eventually joint optimization of generation and recognition. Building upon Stable Video Diffusion, we introduce GenRec, the first unified framework trained with a random-frame conditioning process so as to learn generalized spatial-temporal representations. The resulting framework can naturally supports generation and recognition, and more importantly is robust even when visual inputs contain limited information. Extensive experiments demonstrate the efficacy of GenRec for both recognition and generation. In particular, GenRec achieves competitive recognition performance, offering 75.8% and 87.2% accuracy on SSV2 and K400, respectively. GenRec also performs the best on class-conditioned image-to-video generation, achieving 46.5 and 49.3 FVD scores on SSV2 and EK-100 datasets. Furthermore, GenRec demonstrates extraordinary robustness in scenarios that only limited frames can be observed. Code will be available at https://github.com/wengzejia1/GenRec.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-eaebea0136eb5dc0", "title": "Fundus2Video: Cross-Modal Angiography Video Generation from Static Fundus Photography with Clinical Knowledge Guidance", "authors": ["Weiyi Zhang", "Siyu Huang", "Jiancheng Yang", "Ruoyu Chen", "Zongyuan Ge", "Yingfeng Zheng", "Danli Shi", "Mingguang He"], "year": 2024, "published": "2024-08-27T17:30:49Z", "updated": "2024-08-27T17:30:49Z", "arxiv_id": "http://arxiv.org/abs/2408.15217v1", "url": "https://arxiv.org/abs/2408.15217v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.AI", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Memory / RAG", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Leveraging this mask, our approach integrates innovative knowledge mask-guided techniques, including knowledge-boosted attention, knowledge-aware discriminators, and mask-enhanced patchNCE loss, aimed at refining generation in critical areas and addressing the pixel misalignment challenge.", "Our method achieves the best FVD of 1503.21 and PSNR of 11.81 compared to other common video generation approaches.", "Notably, our knowledge mask surpasses supervised lesion segmentation masks, offering a promising non-invasive alternative to traditional FFA for research and clinical applications."], "abstract": "Fundus Fluorescein Angiography (FFA) is a critical tool for assessing retinal vascular dynamics and aiding in the diagnosis of eye diseases. However, its invasive nature and less accessibility compared to Color Fundus (CF) images pose significant challenges. Current CF to FFA translation methods are limited to static generation. In this work, we pioneer dynamic FFA video generation from static CF images. We introduce an autoregressive GAN for smooth, memory-saving frame-by-frame FFA synthesis. To enhance the focus on dynamic lesion changes in FFA regions, we design a knowledge mask based on clinical experience. Leveraging this mask, our approach integrates innovative knowledge mask-guided techniques, including knowledge-boosted attention, knowledge-aware discriminators, and mask-enhanced patchNCE loss, aimed at refining generation in critical areas and addressing the pixel misalignment challenge. Our method achieves the best FVD of 1503.21 and PSNR of 11.81 compared to other common video generation approaches. Human assessment by an ophthalmologist confirms its high generation quality. Notably, our knowledge mask surpasses supervised lesion segmentation masks, offering a promising non-invasive alternative to traditional FFA for research and clinical applications. The code is available at https://github.com/Michi-3000/Fundus2Video.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7be20b126b2274d1", "title": "Leveraging Hallucinations to Reduce Manual Prompt Dependency in Promptable Segmentation", "authors": ["Jian Hu", "Jiayi Lin", "Junchi Yan", "Shaogang Gong"], "year": 2024, "published": "2024-08-27T17:06:22Z", "updated": "2026-03-09T14:46:34Z", "arxiv_id": "http://arxiv.org/abs/2408.15205v4", "url": "https://arxiv.org/abs/2408.15205v4", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Chain-of-Thought", "Long Context", "Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["While existing methods focus on eliminating hallucinations to improve a model, we argue that MLLM hallucinations can reveal valuable contextual insights when leveraged correctly, as they represent pre-trained large-scale knowledge beyond individual images."], "abstract": "Promptable segmentation typically requires instance-specific manual prompts to guide the segmentation of each desired object. To minimize such a need, task-generic promptable segmentation has been introduced, which employs a single task-generic prompt to segment various images of different objects in the same task. Current methods use Multimodal Large Language Models (MLLMs) to reason detailed instance-specific prompts from a task-generic prompt for improving segmentation accuracy. The effectiveness of this segmentation heavily depends on the precision of these derived prompts. However, MLLMs often suffer hallucinations during reasoning, resulting in inaccurate prompting. While existing methods focus on eliminating hallucinations to improve a model, we argue that MLLM hallucinations can reveal valuable contextual insights when leveraged correctly, as they represent pre-trained large-scale knowledge beyond individual images. In this paper, we utilize hallucinations to mine task-related information from images and verify its accuracy for enhancing precision of the generated prompts. Specifically, we introduce an iterative Prompt-Mask Cycle generation framework (ProMaC) with a prompt generator and a mask generator.The prompt generator uses a multi-scale chain of thought prompting, initially exploring hallucinations for extracting extended contextual knowledge on a test image.These hallucinations are then reduced to formulate precise instance-specific prompts, directing the mask generator to produce masks that are consistent with task semantics by mask semantic alignment. The generated masks iteratively induce the prompt generator to focus more on task-relevant image areas and reduce irrelevant hallucinations, resulting jointly in better prompts and masks. Experiments on 5 benchmarks demonstrate the effectiveness of ProMaC. Code given in https://lwpyh.github.io/ProMaC/.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-97e4778dd4e33461", "title": "X-Reflect: Cross-Reflection Prompting for Multimodal Recommendation", "authors": ["Hanjia Lyu", "Ryan Rossi", "Xiang Chen", "Md Mehrab Tanjim", "Stefano Petrangeli", "Somdeb Sarkhel", "Jiebo Luo"], "year": 2024, "published": "2024-08-27T16:10:21Z", "updated": "2025-10-23T15:44:46Z", "arxiv_id": "http://arxiv.org/abs/2408.15172v2", "url": "https://arxiv.org/abs/2408.15172v2", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.CL", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Extensive experiments conducted on two widely used benchmarks demonstrate that our method outperforms existing prompting baselines in downstream recommendation accuracy.", "To support efficient real-time inference, we also introduce X-Reflect-keyword, a lightweight variant that summarizes image content using keywords and replaces the base model with a smaller backbone, achieving nearly 50% reduction in input length while maintaining competitive performance."], "abstract": "Large Language Models (LLMs) have been shown to enhance the effectiveness of enriching item descriptions, thereby improving the accuracy of recommendation systems. However, most existing approaches either rely on text-only prompting or employ basic multimodal strategies that do not fully exploit the complementary information available from both textual and visual modalities. This paper introduces a novel framework, Cross-Reflection Prompting, termed X-Reflect, designed to address these limitations by prompting Multimodal Large Language Models (MLLMs) to explicitly identify and reconcile supportive and conflicting information between text and images. By capturing nuanced insights from both modalities, this approach generates more comprehensive and contextually rich item representations. Extensive experiments conducted on two widely used benchmarks demonstrate that our method outperforms existing prompting baselines in downstream recommendation accuracy. Furthermore, we identify a U-shaped relationship between text-image dissimilarity and recommendation performance, suggesting the benefit of applying multimodal prompting selectively. To support efficient real-time inference, we also introduce X-Reflect-keyword, a lightweight variant that summarizes image content using keywords and replaces the base model with a smaller backbone, achieving nearly 50% reduction in input length while maintaining competitive performance. This work underscores the importance of integrating multimodal information and presents an effective solution for improving item understanding in multimodal recommendation systems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-f72d7fe44159c857", "title": "Zero-Shot Visual Reasoning by Vision-Language Models: Benchmarking and Analysis", "authors": ["Aishik Nagar", "Shantanu Jaiswal", "Cheston Tan"], "year": 2024, "published": "2024-08-27T14:43:54Z", "updated": "2024-08-27T14:43:54Z", "arxiv_id": "http://arxiv.org/abs/2409.00106v1", "url": "https://arxiv.org/abs/2409.00106v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.CV", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["Chain-of-Thought", "VLM"], "benchmarks": [], "models": ["GPT-3.5-Turbo", "ying"], "key_results": ["In particular, 18% higher accuracy is achieved on the PTR dataset."], "abstract": "Vision-language models (VLMs) have shown impressive zero- and few-shot performance on real-world visual question answering (VQA) benchmarks, alluding to their capabilities as visual reasoning engines. However, the benchmarks being used conflate \"pure\" visual reasoning with world knowledge, and also have questions that involve a limited number of reasoning steps. Thus, it remains unclear whether a VLM's apparent visual reasoning performance is due to its world knowledge, or due to actual visual reasoning capabilities.\n  To clarify this ambiguity, we systematically benchmark and dissect the zero-shot visual reasoning capabilities of VLMs through synthetic datasets that require minimal world knowledge, and allow for analysis over a broad range of reasoning steps. We focus on two novel aspects of zero-shot visual reasoning: i) evaluating the impact of conveying scene information as either visual embeddings or purely textual scene descriptions to the underlying large language model (LLM) of the VLM, and ii) comparing the effectiveness of chain-of-thought prompting to standard prompting for zero-shot visual reasoning.\n  We find that the underlying LLMs, when provided textual scene descriptions, consistently perform better compared to being provided visual embeddings. In particular, 18% higher accuracy is achieved on the PTR dataset. We also find that CoT prompting performs marginally better than standard prompting only for the comparatively large GPT-3.5-Turbo (175B) model, and does worse for smaller-scale models. This suggests the emergence of CoT abilities for visual reasoning in LLMs at larger scales even when world knowledge is limited. Overall, we find limitations in the abilities of VLMs and LLMs for more complex visual reasoning, and highlight the important role that LLMs can play in visual reasoning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-bf3f7d11f252a446", "title": "Negation Blindness in Large Language Models: Unveiling the NO Syndrome in Image Generation", "authors": ["Mohammad Nadeem", "Shahab Saquib Sohail", "Erik Cambria", "Björn W. Schuller", "Amir Hussain"], "year": 2024, "published": "2024-08-27T14:40:16Z", "updated": "2024-09-04T14:40:14Z", "arxiv_id": "http://arxiv.org/abs/2409.00105v2", "url": "https://arxiv.org/abs/2409.00105v2", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["RLVR / GRPO", "Safety / Jailbreak"], "benchmarks": [], "models": ["GPT-4", "Gemini"], "key_results": ["At the same time, researchers are striving to identify the limitations of these tools to improve them further."], "abstract": "Foundational Large Language Models (LLMs) have changed the way we perceive technology. They have been shown to excel in tasks ranging from poem writing and coding to essay generation and puzzle solving. With the incorporation of image generation capability, they have become more comprehensive and versatile AI tools. At the same time, researchers are striving to identify the limitations of these tools to improve them further. Currently identified flaws include hallucination, biases, and bypassing restricted commands to generate harmful content. In the present work, we have identified a fundamental limitation related to the image generation ability of LLMs, and termed it The NO Syndrome. This negation blindness refers to LLMs inability to correctly comprehend NO related natural language prompts to generate the desired images. Interestingly, all tested LLMs including GPT-4, Gemini, and Copilot were found to be suffering from this syndrome. To demonstrate the generalization of this limitation, we carried out simulation experiments and conducted entropy-based and benchmark statistical analysis tests on various LLMs in multiple languages, including English, Hindi, and French. We conclude that the NO syndrome is a significant flaw in current LLMs that needs to be addressed. A related finding of this study showed a consistent discrepancy between image and textual responses as a result of this NO syndrome. We posit that the introduction of a negation context-aware reinforcement learning based feedback loop between the LLMs textual response and generated image could help ensure the generated text is based on both the LLMs correct contextual understanding of the negation query and the generated visual output.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-dbd8058d1a745f90", "title": "MMASD+: A Novel Dataset for Privacy-Preserving Behavior Analysis of Children with Autism Spectrum Disorder", "authors": ["Pavan Uttej Ravva", "Behdokht Kiafar", "Pinar Kullu", "Jicheng Li", "Anjana Bhat", "Roghayeh Leila Barmaki"], "year": 2024, "published": "2024-08-27T14:05:48Z", "updated": "2024-08-28T20:30:29Z", "arxiv_id": "http://arxiv.org/abs/2408.15077v2", "url": "https://arxiv.org/abs/2408.15077v2", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["This framework achieves an accuracy of 95.03% for predicting action types and 96.42% for predicting ASD presence, demonstrating over a 10% improvement compared to models trained on single data modalities."], "abstract": "Autism spectrum disorder (ASD) is characterized by significant challenges in social interaction and comprehending communication signals. Recently, therapeutic interventions for ASD have increasingly utilized Deep learning powered-computer vision techniques to monitor individual progress over time. These models are trained on private, non-public datasets from the autism community, creating challenges in comparing results across different models due to privacy-preserving data-sharing issues. This work introduces MMASD+, an enhanced version of the novel open-source dataset called Multimodal ASD (MMASD). MMASD+ consists of diverse data modalities, including 3D-Skeleton, 3D Body Mesh, and Optical Flow data. It integrates the capabilities of Yolov8 and Deep SORT algorithms to distinguish between the therapist and children, addressing a significant barrier in the original dataset. Additionally, a Multimodal Transformer framework is proposed to predict 11 action types and the presence of ASD. This framework achieves an accuracy of 95.03% for predicting action types and 96.42% for predicting ASD presence, demonstrating over a 10% improvement compared to models trained on single data modalities. These findings highlight the advantages of integrating multiple data modalities within the Multimodal Transformer framework.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7ac23835a2755bdf", "title": "The Benefits of Balance: From Information Projections to Variance Reduction", "authors": ["Lang Liu", "Ronak Mehta", "Soumik Pal", "Zaid Harchaoui"], "year": 2024, "published": "2024-08-27T13:48:15Z", "updated": "2025-02-11T17:47:11Z", "arxiv_id": "http://arxiv.org/abs/2408.15065v2", "url": "https://arxiv.org/abs/2408.15065v2", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG", "math.ST"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": ["MATH"], "models": [], "key_results": ["Furthermore, we describe how various forms of data balancing in contrastive multimodal learning and self-supervised clustering can be better understood, and even improved upon, owing to our variance reduction viewpoint."], "abstract": "Data balancing across multiple modalities and sources appears in various forms in foundation models in machine learning and AI, e.g. in CLIP and DINO. We show that data balancing across modalities and sources actually offers an unsuspected benefit: variance reduction. We present a non-asymptotic statistical bound that quantifies this variance reduction effect and relates it to the eigenvalue decay of Markov operators. Furthermore, we describe how various forms of data balancing in contrastive multimodal learning and self-supervised clustering can be better understood, and even improved upon, owing to our variance reduction viewpoint.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-191d4fd07870733f", "title": "DocLayLLM: An Efficient Multi-modal Extension of Large Language Models for Text-rich Document Understanding", "authors": ["Wenhui Liao", "Jiapeng Wang", "Hongliang Li", "Chengyu Wang", "Jun Huang", "Lianwen Jin"], "year": 2024, "published": "2024-08-27T13:13:38Z", "updated": "2025-03-19T10:05:04Z", "arxiv_id": "http://arxiv.org/abs/2408.15045v3", "url": "https://arxiv.org/abs/2408.15045v3", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Chain-of-Thought", "VLM"], "benchmarks": [], "models": [], "key_results": ["While Multimodal Large Language Models (MLLMs) have achieved fast progress in this domain, existing approaches either demand significant computational resources or struggle with effective multi-modal integration.", "Our DocLayLLM can achieve remarkable performances with lightweight training settings, showcasing its efficiency and effectiveness.", "Experimental results demonstrate that our DocLayLLM outperforms existing OCR-dependent methods and OCR-free competitors."], "abstract": "Text-rich document understanding (TDU) requires comprehensive analysis of documents containing substantial textual content and complex layouts. While Multimodal Large Language Models (MLLMs) have achieved fast progress in this domain, existing approaches either demand significant computational resources or struggle with effective multi-modal integration. In this paper, we introduce DocLayLLM, an efficient multi-modal extension of LLMs specifically designed for TDU. By lightly integrating visual patch tokens and 2D positional tokens into LLMs' input and encoding the document content using the LLMs themselves, we fully take advantage of the document comprehension capability of LLMs and enhance their perception of OCR information. We have also deeply considered the role of chain-of-thought (CoT) and innovatively proposed the techniques of CoT Pre-training and CoT Annealing. Our DocLayLLM can achieve remarkable performances with lightweight training settings, showcasing its efficiency and effectiveness. Experimental results demonstrate that our DocLayLLM outperforms existing OCR-dependent methods and OCR-free competitors. Code and model are available at https://github.com/whlscut/DocLayLLM.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-15876644b30c556c", "title": "NeuralOOD: Improving Out-of-Distribution Generalization Performance with Brain-machine Fusion Learning Framework", "authors": ["Shuangchen Zhao", "Changde Du", "Hui Li", "Huiguang He"], "year": 2024, "published": "2024-08-27T10:54:37Z", "updated": "2024-08-27T10:54:37Z", "arxiv_id": "http://arxiv.org/abs/2408.14950v1", "url": "https://arxiv.org/abs/2408.14950v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": ["ImageNet"], "models": [], "key_results": ["In this paper, we utilize the multimodal learning method to improve the OOD generalization and propose a novel Brain-machine Fusion Learning (BMFL) framework.", "Specially, we employ a pre-trained visual neural encoding model to predict the functional Magnetic Resonance Imaging (fMRI) from visual features which eliminates the need for the fMRI data collection and pre-processing, effectively reduces the workload associated with conventional BMFL methods.", "Moreover, we introduce the Pearson correlation coefficient maximization regularization method into the training process, which improves the fusion capability with better constrains."], "abstract": "Deep Neural Networks (DNNs) have demonstrated exceptional recognition capabilities in traditional computer vision (CV) tasks. However, existing CV models often suffer a significant decrease in accuracy when confronted with out-of-distribution (OOD) data. In contrast to these DNN models, human can maintain a consistently low error rate when facing OOD scenes, partly attributed to the rich prior cognitive knowledge stored in the human brain. Previous OOD generalization researches only focus on the single modal, overlooking the advantages of multimodal learning method. In this paper, we utilize the multimodal learning method to improve the OOD generalization and propose a novel Brain-machine Fusion Learning (BMFL) framework. We adopt the cross-attention mechanism to fuse the visual knowledge from CV model and prior cognitive knowledge from the human brain. Specially, we employ a pre-trained visual neural encoding model to predict the functional Magnetic Resonance Imaging (fMRI) from visual features which eliminates the need for the fMRI data collection and pre-processing, effectively reduces the workload associated with conventional BMFL methods. Furthermore, we construct a brain transformer to facilitate the extraction of knowledge inside the fMRI data. Moreover, we introduce the Pearson correlation coefficient maximization regularization method into the training process, which improves the fusion capability with better constrains. Our model outperforms the DINOv2 and baseline models on the ImageNet-1k validation dataset as well as six curated OOD datasets, showcasing its superior performance in diverse scenarios.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-06b2d383468a5d3b", "title": "Can Transformers Do Enumerative Geometry?", "authors": ["Baran Hashemi", "Roderic G. Corominas", "Alessandro Giacchetto"], "year": 2024, "published": "2024-08-27T09:44:01Z", "updated": "2025-06-05T12:08:50Z", "arxiv_id": "http://arxiv.org/abs/2408.14915v3", "url": "https://arxiv.org/abs/2408.14915v3", "primary_category": "cs.LG", "categories": ["cs.LG", "math.AG"], "pillar": "Multimodal / World Models", "methods": ["Interpretability", "World Model"], "benchmarks": ["MATH"], "models": [], "key_results": ["How can Transformers model and learn enumerative geometry?"], "abstract": "How can Transformers model and learn enumerative geometry? What is a robust procedure for using Transformers in abductive knowledge discovery within a mathematician-machine collaboration? In this work, we introduce a Transformer-based approach to computational enumerative geometry, specifically targeting the computation of $ψ$-class intersection numbers on the moduli space of curves. By reformulating the problem as a continuous optimization task, we compute intersection numbers across a wide value range from $10^{-45}$ to $10^{45}$. To capture the recursive nature inherent in these intersection numbers, we propose the Dynamic Range Activator (DRA), a new activation function that enhances the Transformer's ability to model recursive patterns and handle severe heteroscedasticity. Given precision requirements for computing the intersections, we quantify the uncertainty of the predictions using Conformal Prediction with a dynamic sliding window adaptive to the partitions of equivalent number of marked points. To the best of our knowledge, there has been no prior work on modeling recursive functions with such a high-variance and factorial growth. Beyond simply computing intersection numbers, we explore the enumerative \"world-model\" of Transformers. Our interpretability analysis reveals that the network is implicitly modeling the Virasoro constraints in a purely data-driven manner. Moreover, through abductive hypothesis testing, probing, and causal inference, we uncover evidence of an emergent internal representation of the the large-genus asymptotic of $ψ$-class intersection numbers. These findings suggest that the network internalizes the parameters of the asymptotic closed-form and the polynomiality phenomenon of intersection numbers in a non-linear manner. This opens up new possibilities in inferring asymptotic closed-form expressions directly from limited amount of data.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-c5daa40d5828faa9", "title": "Option Pricing with Stochastic Volatility, Equity Premium, and Interest Rates", "authors": ["Nicole Hao", "Echo Li", "Diep Luong-Le"], "year": 2024, "published": "2024-08-27T21:45:10Z", "updated": "2024-08-27T21:45:10Z", "arxiv_id": "http://arxiv.org/abs/2408.15416v1", "url": "https://arxiv.org/abs/2408.15416v1", "primary_category": "q-fin.MF", "categories": ["q-fin.MF", "math.AP", "q-fin.CP"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": ["MATH"], "models": ["ying", "yielding"], "key_results": ["This paper presents a new model for options pricing."], "abstract": "This paper presents a new model for options pricing. The Black-Scholes-Merton (BSM) model plays an important role in financial options pricing. However, the BSM model assumes that the risk-free interest rate, volatility, and equity premium are constant, which is unrealistic in the real market. To address this, our paper considers the time-varying characteristics of those parameters. Our model integrates elements of the BSM model, the Heston (1993) model for stochastic variance, the Vasicek model (1977) for stochastic interest rates, and the Campbell and Viceira model (1999, 2001) for stochastic equity premium. We derive a linear second-order parabolic PDE and extend our model to encompass fixed-strike Asian options, yielding a new PDE. In the absence of closed-form solutions for any options from our new model, we utilize finite difference methods to approximate prices for European call and up-and-out barrier options, and outline the numerical implementation for fixed-strike Asian call options.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-db4090ce01e991ba", "title": "Evaluating Credit VIX (CDS IV) Prediction Methods with Incremental Batch Learning", "authors": ["Robert Taylor"], "year": 2024, "published": "2024-08-27T20:57:26Z", "updated": "2024-08-27T20:57:26Z", "arxiv_id": "http://arxiv.org/abs/2408.15404v1", "url": "https://arxiv.org/abs/2408.15404v1", "primary_category": "q-fin.CP", "categories": ["q-fin.CP", "cs.LG", "q-fin.RM"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["This paper presents the experimental process and results of SVM, Gradient Boosting, and an Attention-GRU Hybrid model in predicting the Implied Volatility of rolled-over five-year spread contracts of credit default swaps (CDS) on European corporate debt during the quarter following mid-May '24, as represented by the iTraxx/Cboe Europe Main 1-Month Volatility Index (BP Volatility).", "Our comparative assessment aims to identify strengths in SOTA and classical machine learning methods for financial risk prediction"], "abstract": "This paper presents the experimental process and results of SVM, Gradient Boosting, and an Attention-GRU Hybrid model in predicting the Implied Volatility of rolled-over five-year spread contracts of credit default swaps (CDS) on European corporate debt during the quarter following mid-May '24, as represented by the iTraxx/Cboe Europe Main 1-Month Volatility Index (BP Volatility). The analysis employs a feature matrix inspired by Merton's determinants of default probability. Our comparative assessment aims to identify strengths in SOTA and classical machine learning methods for financial risk prediction", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-7ad0f092e87ed870", "title": "Ensuring Equitable Financial Decisions: Leveraging Counterfactual Fairness and Deep Learning for Bias", "authors": ["Saish Shinde"], "year": 2024, "published": "2024-08-27T14:28:06Z", "updated": "2024-08-27T14:28:06Z", "arxiv_id": "http://arxiv.org/abs/2408.16088v1", "url": "https://arxiv.org/abs/2408.16088v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["We show that these approaches are effective in achieving more equitable results through thorough testing and assessment on a skewed financial dataset."], "abstract": "Concerns regarding fairness and bias have been raised in recent years due to the growing use of machine learning models in crucial decision-making processes, especially when it comes to delicate characteristics like gender. In order to address biases in machine learning models, this research paper investigates advanced bias mitigation techniques, with a particular focus on counterfactual fairness in conjunction with data augmentation. The study looks into how these integrated approaches can lessen gender bias in the financial industry, specifically in loan approval procedures. We show that these approaches are effective in achieving more equitable results through thorough testing and assessment on a skewed financial dataset. The findings emphasize how crucial it is to use fairness-aware techniques when creating machine learning models in order to guarantee morally righteous and impartial decision-making.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-b210c30e53e612ec", "title": "Constrained Diffusion Models via Dual Training", "authors": ["Shervin Khalafi", "Dongsheng Ding", "Alejandro Ribeiro"], "year": 2024, "published": "2024-08-27T14:25:42Z", "updated": "2024-11-22T05:41:58Z", "arxiv_id": "http://arxiv.org/abs/2408.15094v2", "url": "https://arxiv.org/abs/2408.15094v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV", "math.OC"], "pillar": "Quant × AI", "methods": ["Video Generation"], "benchmarks": ["MATH"], "models": ["ying"], "key_results": ["We show that our constrained diffusion models generate new data from a mixture data distribution that achieves the optimal trade-off among objective and constraints."], "abstract": "Diffusion models have attained prominence for their ability to synthesize a probability distribution for a given dataset via a diffusion process, enabling the generation of new data points with high fidelity. However, diffusion processes are prone to generating samples that reflect biases in a training dataset. To address this issue, we develop constrained diffusion models by imposing diffusion constraints based on desired distributions that are informed by requirements. Specifically, we cast the training of diffusion models under requirements as a constrained distribution optimization problem that aims to reduce the distribution difference between original and generated data while obeying constraints on the distribution of generated data. We show that our constrained diffusion models generate new data from a mixture data distribution that achieves the optimal trade-off among objective and constraints. To train constrained diffusion models, we develop a dual training algorithm and characterize the optimality of the trained constrained diffusion model. We empirically demonstrate the effectiveness of our constrained models in two constrained generation tasks: (i) we consider a dataset with one or more underrepresented classes where we train the model with constraints to ensure fairly sampling from all classes during inference; (ii) we fine-tune a pre-trained diffusion model to sample from a new dataset while avoiding overfitting.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-ae0b496d3904f86a", "title": "Leveraging RNNs and LSTMs for Synchronization Analysis in the Indian Stock Market: A Threshold-Based Classification Approach", "authors": ["Sanjay Sathish", "Charu C Sharma"], "year": 2024, "published": "2024-08-27T11:08:37Z", "updated": "2024-08-27T11:08:37Z", "arxiv_id": "http://arxiv.org/abs/2409.06728v1", "url": "https://arxiv.org/abs/2409.06728v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "cs.LG"], "pillar": "Quant × AI", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Our research presents a new approach for forecasting the synchronization of stock prices using machine learning and non-linear time-series analysis."], "abstract": "Our research presents a new approach for forecasting the synchronization of stock prices using machine learning and non-linear time-series analysis. To capture the complex non-linear relationships between stock prices, we utilize recurrence plots (RP) and cross-recurrence quantification analysis (CRQA). By transforming Cross Recurrence Plot (CRP) data into a time-series format, we enable the use of Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) networks for predicting stock price synchronization through both regression and classification. We apply this methodology to a dataset of 20 highly capitalized stocks from the Indian market over a 21-year period. The findings reveal that our approach can predict stock price synchronization, with an accuracy of 0.98 and F1 score of 0.83 offering valuable insights for developing effective trading strategies and risk management tools.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-fc0ca2546f05afa8", "title": "Dynamic operator management in meta-heuristics using reinforcement learning: an application to permutation flowshop scheduling problems", "authors": ["Maryam Karimi Mamaghan", "Mehrdad Mohammadi", "Wout Dullaert", "Daniele Vigo", "Amir Pirayesh"], "year": 2024, "published": "2024-08-27T08:38:17Z", "updated": "2024-08-27T08:38:17Z", "arxiv_id": "http://arxiv.org/abs/2408.14864v1", "url": "https://arxiv.org/abs/2408.14864v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["The results demonstrate the superior performance of the proposed framework against state-of-the-art algorithms in terms of optimality gap and convergence speed."], "abstract": "This study develops a framework based on reinforcement learning to dynamically manage a large portfolio of search operators within meta-heuristics. Using the idea of tabu search, the framework allows for continuous adaptation by temporarily excluding less efficient operators and updating the portfolio composition during the search. A Q-learning-based adaptive operator selection mechanism is used to select the most suitable operator from the dynamically updated portfolio at each stage. Unlike traditional approaches, the proposed framework requires no input from the experts regarding the search operators, allowing domain-specific non-experts to effectively use the framework. The performance of the proposed framework is analyzed through an application to the permutation flowshop scheduling problem. The results demonstrate the superior performance of the proposed framework against state-of-the-art algorithms in terms of optimality gap and convergence speed.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-223f613f4445ab09", "title": "MLP, XGBoost, KAN, TDNN, and LSTM-GRU Hybrid RNN with Attention for SPX and NDX European Call Option Pricing", "authors": ["Boris Ter-Avanesov", "Homayoon Beigi"], "year": 2024, "published": "2024-08-26T21:32:49Z", "updated": "2024-10-14T00:46:45Z", "arxiv_id": "http://arxiv.org/abs/2409.06724v3", "url": "https://arxiv.org/abs/2409.06724v3", "primary_category": "q-fin.CP", "categories": ["q-fin.CP", "cs.CE", "cs.LG"], "pillar": "Quant × AI", "methods": ["Deep Research"], "benchmarks": [], "models": [], "key_results": ["In this study, we attempt to leverage the ability of supervised learning methods, such as ANNs, KANs, and gradient-boosted decision trees, to approximate complex multivariate functions in order to calibrate option prices based on past market data.", "Also, the best TDNN model outperforms the best MLP model on all error metrics.", "Also, the KAN model outperforms the TDNN and MLP models."], "abstract": "We explore the performance of various artificial neural network architectures, including a multilayer perceptron (MLP), Kolmogorov-Arnold network (KAN), LSTM-GRU hybrid recursive neural network (RNN) models, and a time-delay neural network (TDNN) for pricing European call options. In this study, we attempt to leverage the ability of supervised learning methods, such as ANNs, KANs, and gradient-boosted decision trees, to approximate complex multivariate functions in order to calibrate option prices based on past market data. The motivation for using ANNs and KANs is the Universal Approximation Theorem and Kolmogorov-Arnold Representation Theorem, respectively. Specifically, we use S\\&P 500 (SPX) and NASDAQ 100 (NDX) index options traded during 2015-2023 with times to maturity ranging from 15 days to over 4 years (OptionMetrics IvyDB US dataset). Black \\& Scholes's (BS) PDE \\cite{Black1973} model's performance in pricing the same options compared to real data is used as a benchmark. This model relies on strong assumptions, and it has been observed and discussed in the literature that real data does not match its predictions. Supervised learning methods are widely used as an alternative for calibrating option prices due to some of the limitations of this model. In our experiments, the BS model underperforms compared to all of the others. Also, the best TDNN model outperforms the best MLP model on all error metrics. We implement a simple self-attention mechanism to enhance the RNN models, significantly improving their performance. The best-performing model overall is the LSTM-GRU hybrid RNN model with attention. Also, the KAN model outperforms the TDNN and MLP models. We analyze the performance of all models by ticker, moneyness category, and over/under/correctly-priced percentage.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-d1752eeefb2c2f2e", "title": "Aiding Humans in Financial Fraud Decision Making: Toward an XAI-Visualization Framework", "authors": ["Angelos Chatzimparmpas", "Evanthia Dimara"], "year": 2024, "published": "2024-08-26T18:10:07Z", "updated": "2024-08-26T18:10:07Z", "arxiv_id": "http://arxiv.org/abs/2408.14552v1", "url": "https://arxiv.org/abs/2408.14552v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CY", "cs.HC"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["AI prevails in financial fraud detection and decision making."], "abstract": "AI prevails in financial fraud detection and decision making. Yet, due to concerns about biased automated decision making or profiling, regulations mandate that final decisions are made by humans. Financial fraud investigators face the challenge of manually synthesizing vast amounts of unstructured information, including AI alerts, transaction histories, social media insights, and governmental laws. Current Visual Analytics (VA) systems primarily support isolated aspects of this process, such as explaining binary AI alerts and visualizing transaction patterns, thus adding yet another layer of information to the overall complexity. In this work, we propose a framework where the VA system supports decision makers throughout all stages of financial fraud investigation, including data collection, information synthesis, and human criteria iteration. We illustrate how VA can claim a central role in AI-aided decision making, ensuring that human judgment remains in control while minimizing potential biases and labor-intensive tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-9900af5ab637b039", "title": "DSTI at LLMs4OL 2024 Task A: Intrinsic versus extrinsic knowledge for type classification", "authors": ["Hanna Abi Akl"], "year": 2024, "published": "2024-08-26T12:50:27Z", "updated": "2024-08-26T12:50:27Z", "arxiv_id": "http://arxiv.org/abs/2408.14236v1", "url": "https://arxiv.org/abs/2408.14236v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["We introduce semantic towers, an extrinsic knowledge representation method, and compare it to intrinsic knowledge in large language models for ontology learning."], "abstract": "We introduce semantic towers, an extrinsic knowledge representation method, and compare it to intrinsic knowledge in large language models for ontology learning. Our experiments show a trade-off between performance and semantic grounding for extrinsic knowledge compared to a fine-tuned model intrinsic knowledge. We report our findings on the Large Language Models for Ontology Learning (LLMs4OL) 2024 challenge.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-af3aaff9a7125d2f", "title": "Integrating the Expected Future in Load Forecasts with Contextually Enhanced Transformer Models", "authors": ["Raffael Theiler", "Leandro Von Krannichfeldt", "Giovanni Sansavini", "Michael F. Howland", "Olga Fink"], "year": 2024, "published": "2024-08-26T12:32:40Z", "updated": "2025-02-21T10:45:05Z", "arxiv_id": "http://arxiv.org/abs/2409.05884v2", "url": "https://arxiv.org/abs/2409.05884v2", "primary_category": "cs.CY", "categories": ["cs.CY", "cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Compared to other state-of-the-art methods, our approach consistently outperforms existing models, underscoring the value of context-aware deep learning techniques in energy forecasting applications."], "abstract": "Accurate and reliable energy forecasting is essential for power grid operators who strive to minimize extreme forecasting errors that pose significant operational challenges and incur high intra-day trading costs. Incorporating planning information -- such as anticipated user behavior, scheduled events or timetables -- provides substantial contextual information to enhance forecast accuracy and reduce the occurrence of large forecasting errors. Existing approaches, however, lack the flexibility to effectively integrate both dynamic, forward-looking contextual inputs and historical data. In this work, we conceptualize forecasting as a combined forecasting-regression task, formulated as a sequence-to-sequence prediction problem, and introduce contextually-enhanced transformer models designed to leverage all contextual information effectively. We demonstrate the effectiveness of our approach through a primary case study on nationwide railway energy consumption forecasting, where integrating contextual information into transformer models, particularly timetable data, resulted in a significant average mean absolute error reduction of 26.6%. An auxiliary case study on building energy forecasting, leveraging planned office occupancy data, further illustrates the generalizability of our method, showing an average reduction of 56.3% in mean absolute error. Compared to other state-of-the-art methods, our approach consistently outperforms existing models, underscoring the value of context-aware deep learning techniques in energy forecasting applications.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-1fa87fc2c6fb759d", "title": "An Item Response Theory-based R Module for Algorithm Portfolio Analysis", "authors": ["Brodie Oldfield", "Sevvandi Kandanaarachchi", "Ziqi Xu", "Mario Andrés Muñoz"], "year": 2024, "published": "2024-08-26T05:31:46Z", "updated": "2024-08-27T04:36:52Z", "arxiv_id": "http://arxiv.org/abs/2408.14025v2", "url": "https://arxiv.org/abs/2408.14025v2", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Experimental evaluation is crucial in AI research, especially for assessing algorithms across diverse tasks."], "abstract": "Experimental evaluation is crucial in AI research, especially for assessing algorithms across diverse tasks. Many studies often evaluate a limited set of algorithms, failing to fully understand their strengths and weaknesses within a comprehensive portfolio. This paper introduces an Item Response Theory (IRT) based analysis tool for algorithm portfolio evaluation called AIRT-Module. Traditionally used in educational psychometrics, IRT models test question difficulty and student ability using responses to test questions. Adapting IRT to algorithm evaluation, the AIRT-Module contains a Shiny web application and the R package airt. AIRT-Module uses algorithm performance measures to compute anomalousness, consistency, and difficulty limits for an algorithm and the difficulty of test instances. The strengths and weaknesses of algorithms are visualised using the difficulty spectrum of the test instances. AIRT-Module offers a detailed understanding of algorithm capabilities across varied test instances, thus enhancing comprehensive AI method assessment. It is available at https://sevvandi.shinyapps.io/AIRT/ .", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-4fe599a946bfdbb7", "title": "LSR-IGRU: Stock Trend Prediction Based on Long Short-Term Relationships and Improved GRU", "authors": ["Peng Zhu", "Yuante Li", "Yifan Hu", "Qinyuan Liu", "Dawei Cheng", "Yuqi Liang"], "year": 2024, "published": "2024-08-26T02:58:37Z", "updated": "2025-05-11T14:36:17Z", "arxiv_id": "http://arxiv.org/abs/2409.08282v3", "url": "https://arxiv.org/abs/2409.08282v3", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "cs.CE", "cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Therefore, we propose a stock price trend prediction model named LSR-IGRU in this paper, which is based on long short-term stock relationships and an improved GRU input.", "Next, we improve the inputs of the GRU model at each step, enabling the model to more effectively integrate temporal information and long short-term relationship information, thereby significantly improving the accuracy of predicting stock trend changes.", "Finally, through extensive experiments on multiple datasets from stock markets in China and the United States, we validate the superiority of the proposed LSR-IGRU model over the current state-of-the-art baseline models."], "abstract": "Stock price prediction is a challenging problem in the field of finance and receives widespread attention. In recent years, with the rapid development of technologies such as deep learning and graph neural networks, more research methods have begun to focus on exploring the interrelationships between stocks. However, existing methods mostly focus on the short-term dynamic relationships of stocks and directly integrating relationship information with temporal information. They often overlook the complex nonlinear dynamic characteristics and potential higher-order interaction relationships among stocks in the stock market. Therefore, we propose a stock price trend prediction model named LSR-IGRU in this paper, which is based on long short-term stock relationships and an improved GRU input. Firstly, we construct a long short-term relationship matrix between stocks, where secondary industry information is employed for the first time to capture long-term relationships of stocks, and overnight price information is utilized to establish short-term relationships. Next, we improve the inputs of the GRU model at each step, enabling the model to more effectively integrate temporal information and long short-term relationship information, thereby significantly improving the accuracy of predicting stock trend changes. Finally, through extensive experiments on multiple datasets from stock markets in China and the United States, we validate the superiority of the proposed LSR-IGRU model over the current state-of-the-art baseline models. We also apply the proposed model to the algorithmic trading system of a financial company, achieving significantly higher cumulative portfolio returns compared to other baseline methods. Our sources are released at https://github.com/ZP1481616577/Baselines_LSR-IGRU.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-2374c69633d4da10", "title": "Generalization of Graph Neural Networks is Robust to Model Mismatch", "authors": ["Zhiyang Wang", "Juan Cervino", "Alejandro Ribeiro"], "year": 2024, "published": "2024-08-25T16:00:44Z", "updated": "2024-09-10T16:28:24Z", "arxiv_id": "http://arxiv.org/abs/2408.13878v2", "url": "https://arxiv.org/abs/2408.13878v2", "primary_category": "cs.LG", "categories": ["cs.LG", "eess.SP"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Graph neural networks (GNNs) have demonstrated their effectiveness in various tasks supported by their generalization capabilities."], "abstract": "Graph neural networks (GNNs) have demonstrated their effectiveness in various tasks supported by their generalization capabilities. However, the current analysis of GNN generalization relies on the assumption that training and testing data are independent and identically distributed (i.i.d). This imposes limitations on the cases where a model mismatch exists when generating testing data. In this paper, we examine GNNs that operate on geometric graphs generated from manifold models, explicitly focusing on scenarios where there is a mismatch between manifold models generating training and testing data. Our analysis reveals the robustness of the GNN generalization in the presence of such model mismatch. This indicates that GNNs trained on graphs generated from a manifold can still generalize well to unseen nodes and graphs generated from a mismatched manifold. We attribute this mismatch to both node feature perturbations and edge perturbations within the generated graph. Our findings indicate that the generalization gap decreases as the number of nodes grows in the training graph while increasing with larger manifold dimension as well as larger mismatch. Importantly, we observe a trade-off between the generalization of GNNs and the capability to discriminate high-frequency components when facing a model mismatch. The most important practical consequence of this analysis is to shed light on the filter design of generalizable GNNs robust to model mismatch. We verify our theoretical findings with experiments on multiple real-world datasets.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-51eb000c45edefb2", "title": "A proof of contribution in blockchain using game theoretical deep learning model", "authors": ["Jin Wang"], "year": 2024, "published": "2024-08-25T12:40:19Z", "updated": "2024-08-25T12:40:19Z", "arxiv_id": "http://arxiv.org/abs/2409.07460v1", "url": "https://arxiv.org/abs/2409.07460v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Extensive experiments have demonstrated that our model reduces latency by 584% compared to the state-of-the-art."], "abstract": "Building elastic and scalable edge resources is an inevitable prerequisite for providing platform-based smart city services. Smart city services are delivered through edge computing to provide low-latency applications. However, edge computing has always faced the challenge of limited resources. A single edge device cannot undertake the various intelligent computations in a smart city, and the large-scale deployment of edge devices from different service providers to build an edge resource platform has become a necessity. Selecting computing power from different service providers is a game-theoretic problem. To incentivize service providers to actively contribute their valuable resources and provide low-latency collaborative computing power, we introduce a game-theoretic deep learning model to reach a consensus among service providers on task scheduling and resource provisioning. Traditional centralized resource management approaches are inefficient and lack credibility, while the introduction of blockchain technology can enable decentralized resource trading and scheduling. We propose a contribution-based proof mechanism to provide the low-latency service of edge computing. The deep learning model consists of dual encoders and a single decoder, where the GNN (Graph Neural Network) encoder processes structured decision action data, and the RNN (Recurrent Neural Network) encoder handles time-series task scheduling data. Extensive experiments have demonstrated that our model reduces latency by 584% compared to the state-of-the-art.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-62c8d55f73163b09", "title": "StockTime: A Time Series Specialized Large Language Model Architecture for Stock Price Prediction", "authors": ["Shengkun Wang", "Taoran Ji", "Linhan Wang", "Yanshen Sun", "Shang-Ching Liu", "Amit Kumar", "Chang-Tien Lu"], "year": 2024, "published": "2024-08-25T00:50:33Z", "updated": "2024-08-25T00:50:33Z", "arxiv_id": "http://arxiv.org/abs/2409.08281v1", "url": "https://arxiv.org/abs/2409.08281v1", "primary_category": "q-fin.ST", "categories": ["q-fin.ST", "cs.AI", "cs.CE", "cs.LG"], "pillar": "Quant × AI", "methods": ["Interpretability", "Memory / RAG", "VLM", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Recently, large language models (LLMs) have brought new ways to improve these predictions.", "Our experiments demonstrate that StockTime outperforms recent LLMs, as it gives more accurate predictions while reducing memory usage and runtime costs."], "abstract": "The stock price prediction task holds a significant role in the financial domain and has been studied for a long time. Recently, large language models (LLMs) have brought new ways to improve these predictions. While recent financial large language models (FinLLMs) have shown considerable progress in financial NLP tasks compared to smaller pre-trained language models (PLMs), challenges persist in stock price forecasting. Firstly, effectively integrating the modalities of time series data and natural language to fully leverage these capabilities remains complex. Secondly, FinLLMs focus more on analysis and interpretability, which can overlook the essential features of time series data. Moreover, due to the abundance of false and redundant information in financial markets, models often produce less accurate predictions when faced with such input data. In this paper, we introduce StockTime, a novel LLM-based architecture designed specifically for stock price data. Unlike recent FinLLMs, StockTime is specifically designed for stock price time series data. It leverages the natural ability of LLMs to predict the next token by treating stock prices as consecutive tokens, extracting textual information such as stock correlations, statistical trends and timestamps directly from these stock prices. StockTime then integrates both textual and time series data into the embedding space. By fusing this multimodal data, StockTime effectively predicts stock prices across arbitrary look-back periods. Our experiments demonstrate that StockTime outperforms recent LLMs, as it gives more accurate predictions while reducing memory usage and runtime costs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-ab8f1e519b42ffb0", "title": "Differentially Private Publication of Electricity Time Series Data in Smart Grids", "authors": ["Sina Shaham", "Gabriel Ghinita", "Bhaskar Krishnamachari", "Cyrus Shahabi"], "year": 2024, "published": "2024-08-24T23:30:09Z", "updated": "2024-08-24T23:30:09Z", "arxiv_id": "http://arxiv.org/abs/2408.16017v1", "url": "https://arxiv.org/abs/2408.16017v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": ["phical"], "key_results": ["We demonstrate through extensive experiments, on both real-world and synthetic datasets, that STPT significantly outperforms existing benchmarks, providing a well-balanced trade-off between data utility and user privacy."], "abstract": "Smart grids are a valuable data source to study consumer behavior and guide energy policy decisions. In particular, time-series of power consumption over geographical areas are essential in deciding the optimal placement of expensive resources (e.g., transformers, storage elements) and their activation schedules. However, publication of such data raises significant privacy issues, as it may reveal sensitive details about personal habits and lifestyles. Differential privacy (DP) is well-suited for sanitization of individual data, but current DP techniques for time series lead to significant loss in utility, due to the existence of temporal correlation between data readings. We introduce {\\em STPT (Spatio-Temporal Private Timeseries)}, a novel method for DP-compliant publication of electricity consumption data that analyzes spatio-temporal attributes and captures both micro and macro patterns by leveraging RNNs. Additionally, it employs a partitioning method for releasing electricity consumption time series based on identified patterns. We demonstrate through extensive experiments, on both real-world and synthetic datasets, that STPT significantly outperforms existing benchmarks, providing a well-balanced trade-off between data utility and user privacy.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-ac182ac84c4cc136", "title": "Submodular Maximization Approaches for Equitable Client Selection in Federated Learning", "authors": ["Andrés Catalino Castillo Jiménez", "Ege C. Kaya", "Lintao Ye", "Abolfazl Hashemi"], "year": 2024, "published": "2024-08-24T22:40:31Z", "updated": "2024-08-27T19:27:07Z", "arxiv_id": "http://arxiv.org/abs/2408.13683v2", "url": "https://arxiv.org/abs/2408.13683v2", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "eess.SP", "eess.SY"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["Both approaches utilize submodular function maximization to achieve more balanced models.", "The efficacy of these methods is demonstrated through extensive evaluations across heterogeneous scenarios, revealing significant improvements in fairness as measured by a client dissimilarity metric."], "abstract": "In a conventional Federated Learning framework, client selection for training typically involves the random sampling of a subset of clients in each iteration. However, this random selection often leads to disparate performance among clients, raising concerns regarding fairness, particularly in applications where equitable outcomes are crucial, such as in medical or financial machine learning tasks. This disparity typically becomes more pronounced with the advent of performance-centric client sampling techniques. This paper introduces two novel methods, namely SUBTRUNC and UNIONFL, designed to address the limitations of random client selection. Both approaches utilize submodular function maximization to achieve more balanced models. By modifying the facility location problem, they aim to mitigate the fairness concerns associated with random selection. SUBTRUNC leverages client loss information to diversify solutions, while UNIONFL relies on historical client selection data to ensure a more equitable performance of the final model. Moreover, these algorithms are accompanied by robust theoretical guarantees regarding convergence under reasonable assumptions. The efficacy of these methods is demonstrated through extensive evaluations across heterogeneous scenarios, revealing significant improvements in fairness as measured by a client dissimilarity metric.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-5b3f083fc077ed56", "title": "EUR-USD Exchange Rate Forecasting Based on Information Fusion with Large Language Models and Deep Learning Methods", "authors": ["Hongcheng Ding", "Xuanze Zhao", "Ruiting Deng", "Shamsul Nahar Abdullah", "Deshinta Arrova Dewi"], "year": 2024, "published": "2024-08-23T16:46:36Z", "updated": "2025-06-27T06:57:32Z", "arxiv_id": "http://arxiv.org/abs/2408.13214v2", "url": "https://arxiv.org/abs/2408.13214v2", "primary_category": "q-fin.CP", "categories": ["q-fin.CP", "cs.AI", "cs.CE", "cs.CL"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["yielding"], "key_results": ["Experiments demonstrate that the proposed method outperforms benchmark models, reducing MAE by 10.69% and RMSE by 9.56% compared to the best performing baseline."], "abstract": "Accurate forecasting of the EUR/USD exchange rate is crucial for investors, businesses, and policymakers. This paper proposes a novel framework, IUS, that integrates unstructured textual data from news and analysis with structured data on exchange rates and financial indicators to enhance exchange rate prediction. The IUS framework employs large language models for sentiment polarity scoring and exchange rate movement classification of texts. These textual features are combined with quantitative features and input into a Causality-Driven Feature Generator. An Optuna-optimized Bi-LSTM model is then used to forecast the EUR/USD exchange rate. Experiments demonstrate that the proposed method outperforms benchmark models, reducing MAE by 10.69% and RMSE by 9.56% compared to the best performing baseline. Results also show the benefits of data fusion, with the combination of unstructured and structured data yielding higher accuracy than structured data alone. Furthermore, feature selection using the top 12 important quantitative features combined with the textual features proves most effective. The proposed IUS framework and Optuna-Bi-LSTM model provide a powerful new approach for exchange rate forecasting through multi-source data integration.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-690e991dc4083420", "title": "Dynamic Label Adversarial Training for Deep Learning Robustness Against Adversarial Attacks", "authors": ["Zhenyu Liu", "Haoran Duan", "Huizhi Liang", "Yang Long", "Vaclav Snasel", "Guiseppe Nicosia", "Rajiv Ranjan", "Varun Ojha"], "year": 2024, "published": "2024-08-23T14:25:12Z", "updated": "2024-08-23T14:25:12Z", "arxiv_id": "http://arxiv.org/abs/2408.13102v1", "url": "https://arxiv.org/abs/2408.13102v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "Quant × AI", "methods": ["Distillation", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Adversarial training is one of the most effective methods for enhancing model robustness."], "abstract": "Adversarial training is one of the most effective methods for enhancing model robustness. Recent approaches incorporate adversarial distillation in adversarial training architectures. However, we notice two scenarios of defense methods that limit their performance: (1) Previous methods primarily use static ground truth for adversarial training, but this often causes robust overfitting; (2) The loss functions are either Mean Squared Error or KL-divergence leading to a sub-optimal performance on clean accuracy. To solve those problems, we propose a dynamic label adversarial training (DYNAT) algorithm that enables the target model to gradually and dynamically gain robustness from the guide model's decisions. Additionally, we found that a budgeted dimension of inner optimization for the target model may contribute to the trade-off between clean accuracy and robust accuracy. Therefore, we propose a novel inner optimization method to be incorporated into the adversarial training. This will enable the target model to adaptively search for adversarial examples based on dynamic labels from the guiding model, contributing to the robustness of the target model. Extensive experiments validate the superior performance of our approach.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-f841341f7af540ad", "title": "Controllable Financial Market Generation with Diffusion Guided Meta Agent", "authors": ["Yu-Hao Huang", "Chang Xu", "Yang Liu", "Weiqing Liu", "Wu-Jun Li", "Jiang Bian"], "year": 2024, "published": "2024-08-23T11:15:36Z", "updated": "2026-01-15T18:50:40Z", "arxiv_id": "http://arxiv.org/abs/2408.12991v3", "url": "https://arxiv.org/abs/2408.12991v3", "primary_category": "cs.CE", "categories": ["cs.CE", "q-fin.TR"], "pillar": "Quant × AI", "methods": ["Video Generation", "Quant / Trading"], "benchmarks": [], "models": ["yield"], "key_results": ["Extensive experimental results show that DigMA achieves superior controllability and generation fidelity."], "abstract": "Generative modeling has transformed many fields, such as language and visual modeling, while its application in financial markets remains under-explored. As the minimal unit within a financial market is an order, order-flow modeling represents a fundamental generative financial task. However, current approaches often yield unsatisfactory fidelity in generating order flow, and their generation lacks controllability, thereby limiting their practical applications. In this paper, we formulate the challenge of controllable financial market generation, and propose a Diffusion Guided Meta Agent (DigMA) model to address it. Specifically, we employ a conditional diffusion model to capture the dynamics of the market state represented by time-evolving distribution parameters of the mid-price return rate and the order arrival rate, and we define a meta agent with financial economic priors to generate orders from the corresponding distributions. Extensive experimental results show that DigMA achieves superior controllability and generation fidelity. Moreover, we validate its effectiveness as a generative environment for downstream high-frequency trading tasks and its computational efficiency.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-18T07:39:53.690619+00:00"}
{"id": "evg-6bf49eb208887b9d", "title": "TokEval: A Tokenizer Evaluation Suite", "authors": ["Clara Meister"], "year": 2026, "published": "2026-08-18T17:52:52Z", "updated": "2026-08-18T17:52:52Z", "arxiv_id": "http://arxiv.org/abs/2608.18062v1", "url": "https://arxiv.org/abs/2608.18062v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["Language model tokenizers are typically selected with minimal evaluation, despite the fact that their design choices directly impact model capabilities."], "abstract": "Language model tokenizers are typically selected with minimal evaluation, despite the fact that their design choices directly impact model capabilities. This can be partly attributed to a limited understanding of which tokenizer properties affect which aspects of downstream performance. We introduce TokEval, a framework of tokenizer evaluation metrics that goes beyond standard measures like fertility and compression rate to capture linguistically and structurally meaningful properties, e.g., UTF-8 character boundary integrity and digit place-value boundary alignment for mathematics. To validate whether these metrics are predictive of downstream model performance, we conduct controlled language model pretraining experiments, varying solely the tokenizers' training data mixture, pretokenization strategy, and training algorithm. We evaluate the resulting models on bits-per-byte (a tokenizer-agnostic version of perplexity) and several benchmarks, spanning linguistic understanding, mathematical reasoning, and code generation. Our experiments suggest that different intrinsic properties have different impacts on model abilities: information-theoretic metrics predict language modeling abilities (Spearman rho up to 0.80), while structure-sensitive metrics, such as those measuring digit and line-break handling, correlate with task accuracy. We hope TokEval enables more principled tokenizer evaluation, replacing pretraining sweeps with intrinsic measurement wherever the two agree.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-ff49eebe97b819e1", "title": "Chain-of-Experience for Continual LLM Improvement", "authors": ["Haoqin Tu", "Yunhao Fang", "Yizhong Wang", "Cihang Xie", "Shen Yan"], "year": 2026, "published": "2026-08-18T17:22:54Z", "updated": "2026-08-18T17:22:54Z", "arxiv_id": "http://arxiv.org/abs/2608.18027v1", "url": "https://arxiv.org/abs/2608.18027v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Test-time Scaling"], "benchmarks": ["MATH"], "models": ["GPT-5", "Claude-4.5 Sonnet. Our study shows that leveraging iterative experience consistently outperforms feedback-free baselines", "Gemini-2.5 Pro", "yields"], "key_results": ["Humans continuously learn from experience, whereas conventional large language model (LLM) evaluations ignore the models' ability to improve through inference-time interaction.", "In this paper, we study how LLMs learn from iterative experience at test time, a setting we refer to as Chain-of-Experience (CoE), where models accumulate experiential traces through iterative interactions with self or environmental feedback to form a continual improvement loop beyond zero-shot inference.", "Our study shows that leveraging iterative experience consistently outperforms feedback-free baselines, achieving substantial gains with self feedback alone, alongside a 5.6% overall improvement and 19% lower API cost across tasks and models."], "abstract": "Humans continuously learn from experience, whereas conventional large language model (LLM) evaluations ignore the models' ability to improve through inference-time interaction. In this paper, we study how LLMs learn from iterative experience at test time, a setting we refer to as Chain-of-Experience (CoE), where models accumulate experiential traces through iterative interactions with self or environmental feedback to form a continual improvement loop beyond zero-shot inference. We instantiate CoE with diverse feedback mechanisms, including model self-feedback and environmental signals such as correctness or public coding test pass rates, and evaluate across math, coding, and knowledge domains using 8 LLMs, including GPT-5, Gemini-2.5 Pro, Claude-4.5 Sonnet. Our study shows that leveraging iterative experience consistently outperforms feedback-free baselines, achieving substantial gains with self feedback alone, alongside a 5.6% overall improvement and 19% lower API cost across tasks and models. We further show that combining complementary feedback channels (e.g., model and correctness signals) yields additional gains, and that CoE delivers higher accuracy per token than existing test-time strategies. We observe a positive correlation between LLM base ability and improvement capacity, and show that models remain robust under weak or spurious feedback, with different feedback contributing to distinct improvement aspects and most gains emerging early in the iterations.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-27e37bf5023f65ec", "title": "Can Large Language Models Explain Flight Safety Events? A Prior-Guided Semantic LLM-based Approach", "authors": ["Lu Xu", "Xu Li", "Linjiang Zheng", "Fan Li", "Riquan Zhang", "Jiaxing Shang"], "year": 2026, "published": "2026-08-18T17:11:40Z", "updated": "2026-08-18T17:11:40Z", "arxiv_id": "http://arxiv.org/abs/2608.18017v1", "url": "https://arxiv.org/abs/2608.18017v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Interpretability", "Safety / Jailbreak"], "benchmarks": [], "models": ["ying"], "key_results": ["In addition, since LLMs are not inherently strong classifiers, CatBoost is incorporated as a statistical expert, and its prediction results are injected into the prompt as prior guidance.", "Experimental results show that the proposed approach achieves competitive classification performance while generating direct and reasonable explanations for event causes."], "abstract": "Improving flight safety with flight data requires not only accurate detection of risk events, but more importantly, clear interpretation of their underlying causes at the level of pilot control behavior. Existing explainable AI techniques, such as feature importance maps, often require considerable domain knowledge to translate them into operationally meaningful explanations. Large Language Models (LLMs), which excel at language reasoning, bring a promising solution to this issue. However, applying LLMs in this domain presents key challenges such as modal inconsistency, limited classification ability, scarcity of task-specific data for fine-tuning, and lack of domain knowledge. To overcome these challenges, we propose FlightLLM, a prior-guided semantic LLM-based approach for interpretable flight safety analysis. Specifically, we first perform feature engineering to address modal inconsistency, combining statistical descriptors with physically meaningful flight indicators. This representation is further processed by a Semantic Discretization module, which converts abstract numerical patterns into qualitative descriptions that are more compatible with language reasoning. In addition, since LLMs are not inherently strong classifiers, CatBoost is incorporated as a statistical expert, and its prediction results are injected into the prompt as prior guidance. A contrastive few-shot learning strategy is further adopted to compensate for limited data. Finally, we design structured prompts to embed aviation-specific knowledge into the inference process. Using hard landing, a representative risk event with complex causal mechanisms, as an anchor point, we evaluate FlightLLM on a dataset of 704 real-world A320 flight samples. Experimental results show that the proposed approach achieves competitive classification performance while generating direct and reasonable explanations for event causes.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-3cf655b8dbc8bc18", "title": "The IOL-AI Challenge: An Open Challenge towards Advancing Linguistic Reasoning", "authors": ["Eduardo Sánchez", "Rita Berrada", "Dan-Mircea Mirea", "Sara Rajaee", "Alexander Piperski", "Ana Meta Dolinar", "Boris Iomdin", "Andrey Nikulin", "Mariya Shmatova", "Marzieh Fadaee", "Julia Kreutzer"], "year": 2026, "published": "2026-08-18T17:00:43Z", "updated": "2026-08-18T17:00:43Z", "arxiv_id": "http://arxiv.org/abs/2608.18011v1", "url": "https://arxiv.org/abs/2608.18011v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["Claude Opus 4.8 earning a jury score equivalent to a gold medal"], "key_results": ["Capability was not determined by scale: 14B submissions outperform models twice their size, and gains come from decoding and output-handling rather than model capacity."], "abstract": "Reasoning in LLMs is overwhelmingly studied in domains that provide a model with rules: mathematics and code. Linguistic puzzles invert this: the solver must first discover the system before reasoning within it. We present the IOL-AI Challenge, an open-science competition run on the unseen problems of the International Linguistics Olympiad (IOL) 2026 Individual Contest, evaluated both automatically and, for the first time, by members of the official IOL Jury under the same rubrics applied to human contestants. The challenge drew 731 submissions from 46 teams under a strict compute budget (one T4, 30 mins). We additionally benchmark 15 unconstrained frontier and open models, with Claude Opus 4.8 earning a jury score equivalent to a gold medal, while both resource-constrained systems we submitted for jury grading scored in the range of the bottom 5% of contestants. Capability was not determined by scale: 14B submissions outperform models twice their size, and gains come from decoding and output-handling rather than model capacity. We also found that automatic metrics rank systems exactly as the jury does, but compress the scale, upscoring weak systems by ~13 points and understating strong ones. Our analysis shows that while frontier models might have prior knowledge about some of the problem languages, it does not significantly help them solve the linguistic reasoning tasks, leaving linguistic reasoning as a strong benchmarking proxy for generalizable reasoning skills.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-a6ab10c2ebeb604a", "title": "Policy-Invariant Reward Shaping from LLM Feedback: A Framework for Hybrid RL Agents", "authors": ["Christophe D. Hounwanou", "John Emeka Eze", "Yaé U. Gaba"], "year": 2026, "published": "2026-08-18T16:55:46Z", "updated": "2026-08-18T16:55:46Z", "arxiv_id": "http://arxiv.org/abs/2608.18008v1", "url": "https://arxiv.org/abs/2608.18008v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["Combining large language models with reinforcement learning is increasingly explored, yet the theoretical status of LLM-derived reward signals is often left implicit."], "abstract": "Combining large language models with reinforcement learning is increasingly explored, yet the theoretical status of LLM-derived reward signals is often left implicit. We formalize the hybrid LLM-planner and RL-controller architecture as a Goal-Augmented Markov Decision Process and show that when the LLM per-state progress score is used as a bounded potential function, the resulting shaping term preserves the optimal policy set even when the LLM scores are inaccurate. This guarantee is stronger than what general LLM-as-reward approaches provide. We verify the result numerically on a small MDP under four potential configurations, including an adversarial one scaled to twenty times the base reward magnitude.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-59d73a6e6a5a72b5", "title": "Composing Flow-Matching Energies with Known Physics: Generation, OOD Detection, and Inversion on PDE Fields", "authors": ["Yixuan Sun", "Anirban Samaddar", "Sandeep Madireddy"], "year": 2026, "published": "2026-08-18T16:53:08Z", "updated": "2026-08-18T16:53:08Z", "arxiv_id": "http://arxiv.org/abs/2608.18004v1", "url": "https://arxiv.org/abs/2608.18004v1", "primary_category": "cs.LG", "categories": ["cs.LG", "physics.comp-ph"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Test-time Scaling"], "benchmarks": [], "models": ["yield", "yields"], "key_results": ["Furthermore, we demonstrate utilizing the data energy and physics-based energy (e.g., PDE residuals) as complementary mechanisms to improve detection accuracy for OOD tasks."], "abstract": "Probabilistic modeling of physical fields benefits from both a data-driven prior and known physical structure such as the governing equations. Energy-based models (EBMs) are a natural fit since energies compose additively, which enables augmenting physics information during inference. However, EBMs have been difficult to train and sample from due to the intractable partition function. We show in this work that flow matching models with a potential-induced velocity yield an explicit scalar energy at all transport times, whose gradient is exactly the converted learned score and which recovers the marginal negative log-density at the population optimum. The time-dependent energy functions are obtained purely from the matching regression objective on an independent linear Gaussian interpolation, without a variational form or additional MCMC steps, and the sampling retains the flow ODE. Access to the energy function from a trained model serves three roles: energy-corrected data generation, energy as a scoring function for out-of-distribution (OOD) detection, and energy compositional posterior sampling for inverse problems. In particular, we show the explicit energy permits general MCMC samplers in the predictor-corrector sampling framework, reducing PDE residual and spectral distance compared to the flow ODE baseline. Furthermore, we demonstrate utilizing the data energy and physics-based energy (e.g., PDE residuals) as complementary mechanisms to improve detection accuracy for OOD tasks. In addition, we explore the connection to MCMC-based inference for inverse problems by composing the energy with a quadratic observational likelihood that yields a posterior energy, used as an explicitly chosen family of inference-time targets.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-8f192eb8fd702b36", "title": "Recirculation", "authors": ["Michael C. Mozer", "Shoaib Ahmed Siddiqui", "Danny Sawyer", "Sunny Sanyal", "Rosanne Liu"], "year": 2026, "published": "2026-08-18T16:30:21Z", "updated": "2026-08-18T16:30:21Z", "arxiv_id": "http://arxiv.org/abs/2608.17981v1", "url": "https://arxiv.org/abs/2608.17981v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Test-time Scaling"], "benchmarks": ["GSM8K"], "models": ["Gemma3 family"], "key_results": ["We describe an inference-time architectural enhancement for off-the-shelf foundation models that markedly reduces perplexity and boosts accuracy across generation and reasoning tasks.", "Relative to the off-the-shelf baseline, adaptive recirculation achieves remarkable gains on the Gemma3 family, including a 23% reduction in perplexity on a suite of datasets, a 21% increase in accuracy on GSM8k, and reliable improvements in accuracy on other downstream tasks."], "abstract": "We describe an inference-time architectural enhancement for off-the-shelf foundation models that markedly reduces perplexity and boosts accuracy across generation and reasoning tasks. Our approach incurs essentially no additional latency during generation, though it requires serial processing in the prefill phase. Motivated by the fundamental limitation that state updates in feedforward transformers are bounded by model depth, our technique, recirculation, introduces a specific form of recurrence that allows the model to act as a dynamical system and track belief states. We distinguish this technique from chain-of-thought computation---which is better reserved for complex inferences rather than basic state tracking---as well as from popular depth-recurrence techniques (looping) and the costly training of recurrent transformers. We also propose and evaluate an adaptive variant of recirculation which requires only light tuning of hyperparameters while freezing the original model weights. Relative to the off-the-shelf baseline, adaptive recirculation achieves remarkable gains on the Gemma3 family, including a 23% reduction in perplexity on a suite of datasets, a 21% increase in accuracy on GSM8k, and reliable improvements in accuracy on other downstream tasks. Our training-free approach succeeds by leveraging the model itself to inform architectural modifications, suggesting a route to architectural evolution guided by a trained network's properties rather than forced, arbitrary design choices.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-ae3d62ea089f9669", "title": "Towards Zero-Shot Task Transfer with Neurosymbolic World Models", "authors": ["Isidoro Tamassia", "Lennert De Smet", "Giuseppe Marra"], "year": 2026, "published": "2026-08-18T16:12:40Z", "updated": "2026-08-18T16:12:40Z", "arxiv_id": "http://arxiv.org/abs/2608.17959v1", "url": "https://arxiv.org/abs/2608.17959v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Interpretability", "World Model"], "benchmarks": [], "models": ["ying"], "key_results": ["State-of-the-art model-based reinforcement learning methods learn neural world models that allow policy improvement by planning in a latent space, without assumptions on the structure of the underlying environment."], "abstract": "State-of-the-art model-based reinforcement learning methods learn neural world models that allow policy improvement by planning in a latent space, without assumptions on the structure of the underlying environment. While expressive, these models are generally task-dependent: they learn uninterpretable latent representations that are tied to the training task and thus hard to generalize to new tasks. In this work, we present a novel world model formulation where the reward prediction only depends on a subset of structured, symbolic components of the whole latent state. Decoupling observation reconstruction and reward prediction allows us to learn world models that can adapt zero-shot, i.e. without further environment interactions, to new reward functions defined over the same symbolic state space. We discuss the main advantages and challenges of learning these neurosymbolic world models and demonstrate the strong generalisation properties of our approach over purely neural methods.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-bd54f72f1521f43a", "title": "Understanding the Surprising Generalization Properties of Tabular Foundation Models", "authors": ["Nour Shaheen", "Junwei Ma", "Alex Labach", "Frank Hutter", "Valentin Thomas", "Anthony L. Caterini"], "year": 2026, "published": "2026-08-18T16:11:08Z", "updated": "2026-08-18T16:11:08Z", "arxiv_id": "http://arxiv.org/abs/2608.17957v1", "url": "https://arxiv.org/abs/2608.17957v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["We show that the same task-centric perspective can help corpus design at scale: fine-grained column-level pre-processing consistently improves downstream performance, while no improvements are observed when we filter or deduplicate at the dataset level."], "abstract": "Tabular Foundation Models (TFMs) increasingly rely on in-context learning, where a model receives labelled examples at inference time and predicts labels for new inputs without updating its weights. Existing TFMs are typically trained on either massive synthetic corpora or very large collections of real datasets. In contrast, we show that surprisingly strong transfer can emerge from self-supervised pre-training on just a single real table. In this setting, we also find that tables tend to be either broadly useful or broadly poor regardless of downstream prediction task, and that the strongest predictor of usefulness is the number of features rather than the number of instances. This leads to a task-centric interpretation of tabular pre-training: the number and the quality of tasks are essential for the pre-training of TFMs.\n  We show that the same task-centric perspective can help corpus design at scale: fine-grained column-level pre-processing consistently improves downstream performance, while no improvements are observed when we filter or deduplicate at the dataset level.\n  Finally, we offer a new perspective for how TFMs generalize: we believe that tabular in-context generalization is largely retrieval-based, and good models are those that learn to identify relevant examples in the provided context and aggregate them well. The mechanics of TFMs have been relatively understudied; our task-centric, retrieval-based perspective offers a new framework to guide future model and corpus design.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-410715d3c43f2744", "title": "Do Large Language Models Play Six Degrees of Separation? Measuring Topological Compression in Long-Context Manifolds", "authors": ["Md. Faiyaz Abdullah Sayeedi"], "year": 2026, "published": "2026-08-18T16:05:25Z", "updated": "2026-08-18T16:05:25Z", "arxiv_id": "http://arxiv.org/abs/2608.17950v1", "url": "https://arxiv.org/abs/2608.17950v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Long Context", "Interpretability", "Memory / RAG"], "benchmarks": [], "models": ["ying"], "key_results": ["Large Language Models (LLMs) demonstrate remarkable multi-hop reasoning capabilities over long contexts, yet the internal mechanisms enabling these distant cognitive leaps remain poorly understood."], "abstract": "Large Language Models (LLMs) demonstrate remarkable multi-hop reasoning capabilities over long contexts, yet the internal mechanisms enabling these distant cognitive leaps remain poorly understood. Traditional attention-based interpretability often fails to capture true semantic proximity due to routing artifacts like attention sinks. In this paper, we bypass attention weights to directly analyze the dynamic geometry of the hidden state manifold, proving that deep LLM latent spaces natively organize into Small-World networks. By sparsifying the continuous similarity matrices of long-context representations into unweighted graphs, we trace the connectivity between highly disjoint semantic anchors across two distinct architectures. Our findings reveal a sharp topological phase transition: while early syntactic layers remain entirely fractured, deep reasoning layers abruptly compress massive conceptual distances into highly navigable pathways strictly bounded by the \"Six Degrees of Separation\" limit (=< 6 semantic hops). Furthermore, we demonstrate the practical efficacy of this framework by applying it to zero-shot hallucination detection within Retrieval-Augmented Generation (RAG) using the RAGognize dataset. We show that factually grounded generations maintain structural integrity with their source context (approximately 3 hops), whereas hallucinations induce severe topological collapse. Ultimately, this work mathematically formalizes how transformers execute abstract reasoning and provides a novel, strictly geometric signature for evaluating factual reliability.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-39ab3fc1ba0d25d4", "title": "Efficient RLVR Scheduling via Graph-Structured Online Difficulty Estimation", "authors": ["Zhizhao Liu", "Zhiliang Tian", "Xi Wang", "Zhihua Wen", "Yihang Xiong", "Zhiquan Lai", "Dongsheng Li"], "year": 2026, "published": "2026-08-18T16:01:00Z", "updated": "2026-08-18T16:01:00Z", "arxiv_id": "http://arxiv.org/abs/2608.17941v1", "url": "https://arxiv.org/abs/2608.17941v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Interpretability"], "benchmarks": [], "models": [], "key_results": ["Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models but relies on costly rollout exploration.", "Experiments across multiple base models, RL schedulers, and benchmarks demonstrate that our framework achieves better performance."], "abstract": "Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models but relies on costly rollout exploration. Assigning the same exploration budget to samples with different difficulty levels is inefficient: easy samples may receive redundant rollouts, whereas difficult but learnable samples may receive too little exploration. Existing adaptive schedulers address this mismatch through curriculum-based sample selection or non-uniform rollout allocation based on estimated sample difficulty. However, obtaining reliable online difficulty estimates remains challenging: dedicated probing adds substantial generation overhead, whereas history-based estimators face a cold start with no initial observations and stale feedback, and typically ignore relations among samples. To address these limitations, we propose a plug-and-play graph-based online difficulty estimator that shares rollout feedback across related samples and continuously updates their difficulty estimates, mitigating cold start and staleness without dedicated probing. Specifically, we first construct a difficulty-aware sample graph based on semantic and reasoning similarities. Based on this graph, we introduce latent difficulty states and use a Potts prior to encourage neighboring samples to share the same state. We then employ a state-level Beta-Binomial model to aggregate the rollout outcomes associated with each state. Finally, we use an online mean-field variational algorithm to continuously update the latent-state assignments and state-level difficulty as new feedback arrives. Our framework can be integrated into sample-selection and rollout-allocation schedulers, enabling difficulty-adaptive exploration without dedicated probing. Experiments across multiple base models, RL schedulers, and benchmarks demonstrate that our framework achieves better performance.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-bd64b2cd3fdfe5e0", "title": "Grading Needs a Rubric, Not Intelligence", "authors": ["Jhen-Ke Lin"], "year": 2026, "published": "2026-08-18T16:00:05Z", "updated": "2026-08-18T16:00:05Z", "arxiv_id": "http://arxiv.org/abs/2608.17938v1", "url": "https://arxiv.org/abs/2608.17938v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["yielding"], "key_results": ["Small language models can grade open-ended examination answers as reliably as substantially more expensive models when they grade against an explicit rubric."], "abstract": "Small language models can grade open-ended examination answers as reliably as substantially more expensive models when they grade against an explicit rubric. We test this claim as the design principle behind any-to-bench: a frontier model reads source documents once, at ingestion, to extract each question and its rubric; lower-cost models then perform all repeated grading work. We evaluate six cost-efficient model configurations from two model families at three reasoning-effort levels. Each configuration answers 24 open-ended examination questions, and each also grades every answer sheet three times, yielding 3,456 per-question grades. Scores depend overwhelmingly on the answer being graded: answer identity explains 95.6% of score variance, whereas judge identity explains only 0.2%. Raising a writer's reasoning effort moves earned scores by as much as 0.143 of full marks, while raising a judge's reasoning effort moves assigned scores by at most 0.006. Six frontier-tier judges, added as a check, reproduce these scores and are no more reliable as a panel. Two ablations then decompose the rubric on the same questions and answers. Removing its criteria and levels while keeping the official answer changes nothing measurable. Removing the official answer as well collapses reliability (ICC 0.888 to 0.628), inflates scores, and makes judge reasoning effort matter again. The rubric is what decouples grading from judge intelligence, and within the rubric the official answer does nearly all the work. We find no evidence of length preference or same-family preference under rubric-anchored grading.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-82fb5e912e423ee6", "title": "Collective Counterfactual Planning: Coordination, Consent, and Verification under Representational Constraints", "authors": ["Chainarong Amornbunchornvej"], "year": 2026, "published": "2026-08-18T15:52:12Z", "updated": "2026-08-18T15:52:12Z", "arxiv_id": "http://arxiv.org/abs/2608.17932v1", "url": "https://arxiv.org/abs/2608.17932v1", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Groups routinely complete projects that no single member can plan, execute, or verify alone."], "abstract": "Groups routinely complete projects that no single member can plan, execute, or verify alone. We propose a formal model of this phenomenon, Collective Counterfactual Planning (CCP), in which the binding limitation on each agent is neither capability, knowledge, nor observability, but representational geometry: each agent perceives the state, conceives moves, consents to actions, and certifies goal requirements only through a projection onto an agent-specific subspace of a common task space. Four gates jointly determine whether a team can reach a conjunctive goal and legitimately recognize that it has done so: the exogenous implementation coalitions required to perform each action, together with three representational gates -- conception, consent, and task-relative verification qualification. We define the Collective Counterfactual Solvability (CCS) problem, separating geometric feasibility, executable attainment, and validated completion. The results expose a positive-negative duality. Iterated cross-agent relay can unlock a solution that no one-shot pooling of individual plans contains, but any goal requirement depending essentially on the subspace dark to the entire team is unverifiable and therefore not validly completable, even when the trajectory accidentally attains it. Memoryless and audited consent further constrain different objects -- action directions versus cumulative trajectory states -- and neither dominates the other. A four-step exhaustive horizon-bounded solvability scheme is sound and complete under exact representation of the relay closure; restricted implementations remain sound on returned plans but need not be complete. The model gives one geometry for sequential mutual enabling, competent execution of steps whose purpose is invisible to the executor, forced sub-teaming at expertise boundaries, and completion that cannot be validly declared.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-2265717d75b18aa7", "title": "Comparative Study of Out-of-the-Box Technology for Automatic Target Detection and Recognition", "authors": ["Alma M. Liezenga", "Lotte Nijskens", "Henrik R. Baumann", "Stefan Becker", "Simon Bensberg", "Niccolò Camarlinghi", "Håvard R. Eiring", "Alexander W. Johnsgaard", "Tanel Liiv", "Giuseppe Martino", "Matteo Marturini", "Matthias Rapp", "Jan Erik van Woerden", "Alexander Wolpert", "Hugo J. Kuijf"], "year": 2026, "published": "2026-08-18T15:43:34Z", "updated": "2026-08-18T15:43:34Z", "arxiv_id": "http://arxiv.org/abs/2608.17917v1", "url": "https://arxiv.org/abs/2608.17917v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["Recent advances in object detection and artificial intelligence (AI) significantly boosted the potential performance of ATD/R.", "As a solution, this paper explores the use of publicly available models and civilian datasets to achieve reasonable performance in military contexts.", "We benchmark several state-of-the-art models, including six iterations of the YOLO series and two variations on the DETR framework, on a newly acquired military relevant dataset."], "abstract": "Automatic Target Detection and Recognition (ATD/R) is critical for military decision support and (semi-)autonomous operations. Recent advances in object detection and artificial intelligence (AI) significantly boosted the potential performance of ATD/R. However, the scarcity of publicly available military datasets limits the application of these systems. As a solution, this paper explores the use of publicly available models and civilian datasets to achieve reasonable performance in military contexts. We benchmark several state-of-the-art models, including six iterations of the YOLO series and two variations on the DETR framework, on a newly acquired military relevant dataset. This dataset features military vehicles and challenging circumstances, including various degrees of occlusions and small targets. The out-of-the-box version of each model is validated alongside a version finetuned on the VisDrone dataset. This dataset features small objects, an Air-to-Ground (A2G) perspective and relevant classes, potentially generalizing to our military ATD/R task. We compare the performance of the models using mAP@0.5 and mAP@0.5:0.95, across A2G and Ground-to-Ground (G2G) perspective, target size and model size, giving insight into the real-time capabilities of models. Our main findings are: (1) bigger models outperform smaller models, (2) DETR-based models show promising results compared to the YOLO series,(3) fine-tuning models on an out-of-domain A2G dataset, improves their A2G performance and slightly improves their performance on small objects, but (4) all models still struggle with detecting small objects in an A2G scenario. We conclude that, despite recent advances in object detection, in-domain training is still crucial for creating capable ATD/R systems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-5a50cd46503b235d", "title": "CABLE: Extending the Reach of Memory Retrieval via Complementary Antecedent-Based Linking and Expansion", "authors": ["Zheling Tan", "Jin Gao", "Dequan Wang"], "year": 2026, "published": "2026-08-18T15:40:29Z", "updated": "2026-08-18T15:40:29Z", "arxiv_id": "http://arxiv.org/abs/2608.17911v1", "url": "https://arxiv.org/abs/2608.17911v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Memory / RAG"], "benchmarks": [], "models": ["GPT-4o-mini.", "Qwen3.5-27B", "DeepSeek-chat", "yields"], "key_results": ["CABLE yields higher mean LLM-judge scores in every evaluated system-level setting, with the largest gains in categories where useful evidence is distributed across memories or sessions, including open-domain, multi-session, and preference-oriented questions."], "abstract": "As LLM agents operate across structured workflows and sessions, preserving long-term history does not ensure that later contexts can recover relevant evidence through a bounded memory interface. We study this evidence-reachability problem in long-term conversational memory, where retrieval still relies heavily on semantic similarity. This works well for topical recall, but it often misses earlier experiences, plans, or motivations that are semantically distant from the later events they help explain. Existing memory graphs provide cross-memory structure, yet links driven mainly by semantic overlap can duplicate what the host retriever already recovers. We argue that link construction should instead prioritize a sparse set of retriever-complementary associations. We present CABLE (Complementary Antecedent-Based Linking and Expansion), a plug-in augmentation that constructs links designed to extend the host retriever's direct semantic reach. For each new memory, CABLE generates antecedent-oriented queries, retrieves prior memories, subtracts candidates in the direct semantic neighborhood, and verifies the remainder before adding the accepted complementary associations into a sparse directed graph. At retrieval time, CABLE expands the host system's retrieved seeds along these links to surface implicit supporting evidence. We evaluate CABLE with A-MEM on LoCoMo and MA-LongMemEval, and further integrate it into SimpleMem and Mem0g on LoCoMo, using Qwen3.5-27B, DeepSeek-chat, and GPT-4o-mini. CABLE yields higher mean LLM-judge scores in every evaluated system-level setting, with the largest gains in categories where useful evidence is distributed across memories or sessions, including open-domain, multi-session, and preference-oriented questions. These results support prioritizing sparse, reasoning-relevant associations that complement rather than duplicate the host retriever.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-2be0bee6676c99cf", "title": "Multi-Agent AI System for Radiology Report Structuring and Quality Assurance with Independent Radiologist Evaluation", "authors": ["Iryna Hartsock", "Cesar Lam", "Christopher Otteni", "Aliya Qayyum", "Robert Gatenby", "Cyrillo Araujo", "Ghulam Rasool"], "year": 2026, "published": "2026-08-18T17:57:08Z", "updated": "2026-08-18T17:57:08Z", "arxiv_id": "http://arxiv.org/abs/2608.18072v1", "url": "https://arxiv.org/abs/2608.18072v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["It also detected mismatches between the Findings and Impression sections, or within sections; gender-anatomy conflicts; and undocumented communication of critical findings.", "The system flagged 90 (14.1%) reports, most commonly for section mismatches (80 reports, 12.5%)."], "abstract": "Purpose: To develop and evaluate a locally deployed multi-agent AI system for radiology report structuring and quality assurance. Materials and Methods: This retrospective study included 638 radiology reports from CT examinations of the chest, abdomen, and pelvis dictated by 15 board-certified radiologists in 2023 and 2024. A multi-agent AI pipeline was developed to perform report structuring and quality assurance (QA). The system structured the report into standardized anatomical sections at the sentence level using regex rules and local large language models. It also detected mismatches between the Findings and Impression sections, or within sections; gender-anatomy conflicts; and undocumented communication of critical findings. Two board-certified radiologists independently evaluated a 45-report subset. Results: The multi-agent system structured the Findings sections of all reports (22,270 sentences) into a predefined anatomical format while retaining the original report content. The system flagged 90 (14.1%) reports, most commonly for section mismatches (80 reports, 12.5%). In the radiologist evaluation, both reviewers agreed that 31 (69%) were correctly restructured, 2 reports (4%) were incorrectly restructured, and disagreed on the remaining 12 reports (27%). Both reviewers agreed that no clinically important information was omitted and no fabricated content was introduced. Overall QA performance was rated as \"excellent\" or \"good\" in 84% of the evaluated reports, with the remaining reports rated as \"fair\". Conclusion: A locally deployed multi-agent AI system combined radiology report structuring and quality assurance within a single workflow. The system demonstrated favorable performance in radiologist evaluation. Such systems may support standardization of reporting and quality assurance in radiology practice.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-cec159e593b6ac53", "title": "On the Fragility of Self-Improving Agents: Variance, Task Order, and Underspecification", "authors": ["Qinyuan Ye", "Yu Li", "Yada Pruksachatkun", "Jiaxin Zhang", "Chien-Sheng Wu"], "year": 2026, "published": "2026-08-18T17:55:07Z", "updated": "2026-08-18T17:55:07Z", "arxiv_id": "http://arxiv.org/abs/2608.18066v1", "url": "https://arxiv.org/abs/2608.18066v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.LG"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG", "Deep Research"], "benchmarks": [], "models": [], "key_results": ["Memory-based self-improving agents--those that learn from an online stream of tasks and improve over time by maintaining a textual memory bank--have shown great promise in recent literature.", "Second, the agent's improvement is highly dependent on task order."], "abstract": "Memory-based self-improving agents--those that learn from an online stream of tasks and improve over time by maintaining a textual memory bank--have shown great promise in recent literature. However, the reliability aspects of these methods have been critically overlooked. In this work, we conduct a comprehensive re-evaluation of two memory-based methods, broadening the scope of evaluation along two axes: (1) including multiple runs to quantify variance, and (2) randomly shuffling the tasks to investigate the effect of task order. Through these experiments, we make two observations that expose the fragility of current methods: First, agent evaluation is inherently noisy in complex environments and on multi-step tasks, and stacking a self-improving loop on top can further amplify this noise. Second, the agent's improvement is highly dependent on task order. Prior works often adopt default orderings that impose an implicit curriculum, acting as a hidden prerequisite for success.\n  To better understand this fragility, we manually examine the agents' memory and hypothesize that task and environment underspecification contribute to this fragility. We validate this hypothesis by incorporating information that enables better specification, such as detailed rubrics and environment feedback, into the memory construction process. While this added information partially closes the performance degradation in previous experiments, significant gaps still remain, suggesting that other uncharacterized factors contribute to this fragility. Looking ahead, our work advocates for more rigorous evaluation protocols for self-improving agents by reporting results across multiple runs and stress-testing them under challenging conditions. Moreover, our findings on underspecification call for systems and interfaces that enable effective human oversight, preventing agents from failing in unforeseeable ways.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-be3261771ff30deb", "title": "Delegation Asymmetry in Agentic Recommender Systems: Measuring Two-Sided Receptivity in Online Dating", "authors": ["Daria Leshchikova", "Valentina V. Kuskova", "Dmitry Zaytsev", "Valerii Klimov"], "year": 2026, "published": "2026-08-18T17:51:16Z", "updated": "2026-08-18T17:51:16Z", "arxiv_id": "http://arxiv.org/abs/2608.18058v1", "url": "https://arxiv.org/abs/2608.18058v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["ying"], "key_results": ["Autonomous LLM agents that converse on a user's behalf are an emerging design pattern in matching platforms, yet their viability depends on a condition rarely examined: users must accept not only delegating conversation to an agent, but also receiving agent-mediated communication from others."], "abstract": "Autonomous LLM agents that converse on a user's behalf are an emerging design pattern in matching platforms, yet their viability depends on a condition rarely examined: users must accept not only delegating conversation to an agent, but also receiving agent-mediated communication from others. We study this condition using two large-scale surveys of active users of a major dating platform (N=2,894 on generative profile features; N=2,617 on autonomous conversational agents, fielded in two languages). We develop a latent-variable measurement model of agent receptivity based on graded response models with latent regression, and show via model comparison that willingness to send and willingness to receive agent communication are distinct constructs: highly correlated (rho=0.92) but separable (Delta BIC=52), with partial measurement invariance across languages. The model quantifies a systematic delegation asymmetry: deploying one's own agent requires far lower receptivity (threshold -0.38) than engaging a counterpart's agent (+0.32; full engagement +1.39), and mean deployment propensity exceeds engagement propensity roughly threefold. Under a random-pairing counterfactual derived from stated receptivity, only 4-13% of directed dyads combine agent deployment with receiver engagement, with a pronounced gender-directional imbalance. Design counterfactuals quantify the levers: a reciprocity requirement cuts interaction volume by half or more by excluding nearly two-thirds of would-be deployment, while routing agent contacts on receive receptivity triples per-contact engagement, a lift that survives out-of-sample validation with the target item held out (AUC 0.88, 3.1x quartile lift under respondent-level cross-validation). We discuss implications for agentic recommender design, including disclosure, opt-in mechanics, and receptivity-aware matchmaking.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-e23f4e0e6e5dd178", "title": "StagedWorkspace: A Versioned Workspace for Knowledge-Work Agents", "authors": ["Yining Hua", "Hongbin Na", "Yifan Zhou", "Akshay Kalose", "Cyrus Ayubcha", "Levi Lian"], "year": 2026, "published": "2026-08-18T17:44:18Z", "updated": "2026-08-18T17:44:18Z", "arxiv_id": "http://arxiv.org/abs/2608.18050v1", "url": "https://arxiv.org/abs/2608.18050v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["GPT-5.4", "Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX"], "key_results": ["In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access has the highest point estimate for every tested model; relative to the more limiting single view, it improves OfficeQA Pass@1 by 8.3-12.1 points and APEX mean rubric score by 4.7-9.2 points."], "abstract": "AI agents increasingly perform knowledge work (i.e., produce and modify persistent digital artifacts such as code repositories, documents, spreadsheets, slides, reports), yet the parsed views they search, the native files they edit, the changes they review, and the artifacts they submit can refer to different versions of the same work product. We formulate this as a workspace-state contract: every view should be explicitly tied to a version of the evolving workspace state. Coding agents partly address this need through repository contracts for search, diffs, and tests, whereas an analogous contract is less explicit for PDFs, spreadsheets, slides, notebooks, and mixed-format project folders. We propose StagedWorkspace, a versioned workspace for knowledge-work agents. The workspace binds parsed records and review diffs to content hashes of the native files as they change. In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access has the highest point estimate for every tested model; relative to the more limiting single view, it improves OfficeQA Pass@1 by 8.3-12.1 points and APEX mean rubric score by 4.7-9.2 points. SW-AGENT scores 63.9% with Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX, compared with published same-model scores of 29.3% and 25.5, respectively. A paired review-axis ablation on 57 file-editing tasks further finds higher observed scores when diffs are visible. These results identify workspace state as an experimental variable in knowledge-work agents and motivate benchmarks that score evidence, staged edits, and submitted artifacts as explicit state transitions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-13b675957f64d6e6", "title": "Language Has Two Parameters: Narrative-Induced Semantic Plasticity and Phase-Sensitive Interpretation", "authors": ["Hollis Robbins"], "year": 2026, "published": "2026-08-18T17:34:25Z", "updated": "2026-08-18T17:34:25Z", "arxiv_id": "http://arxiv.org/abs/2608.18041v1", "url": "https://arxiv.org/abs/2608.18041v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Interpretability"], "benchmarks": [], "models": [], "key_results": ["The standard transformer has no explicit representation for phase in frozen inference, and the interpretability program measuring progress by monosemanticity is optimizing against it: the coexistence it treats as a defect is the condition of allusion, irony, and quotation."], "abstract": "Language has two parameters. Count how often words occur together and you estimate amplitude, the strength of association. Word embeddings and attention weights refine that count, which sums every writer in the corpus together. This paper claims a second parameter, phase, which signed weights learned from a corpus do not supply. Phase exists only between meanings: it determines how coactivated meanings combine, and it can reverse what a meaning contributes while that meaning stays fully present. A speaker can set phase in the signal through linguistic form; encounters install phase relations and history distributes them. Population averaging deletes history-indexed phase: agent-deindexed corpora identify the population marginal state and determine no individual or dyadic state, at any scale. The standard transformer has no explicit representation for phase in frozen inference, and the interpretability program measuring progress by monosemanticity is optimizing against it: the coexistence it treats as a defect is the condition of allusion, irony, and quotation. Six predictions test whether a suppressed meaning stays active, whether encounter order changes what a phrase does, whether marking the signal changes how a shared phrase is taken, and whether a model given a history is changed by it or only informed about it. The claim defended is the weak version: interpretation requires a second relational parameter, signed, persistent, and indexed to individuals and dyads. Quantum probability is one notation for the parameter; nothing in the formalism claims quantum processes in the brain. The strong version, that the quantum calculus constrains these phenomena as signed classical models do not, rests on an encounter-order constraint not yet derived. The architecture the theory calls for is a language model with agent-indexed, phase-bearing semantic states.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-d4ce7b31c35015f5", "title": "EvoTS-Agent: A Self-Evolving LLM Agent for Financial Time Series Change Point Detection", "authors": ["Lei Jiang", "Ye Wei", "Xinyu Xi", "Jordan Langham-Lopez", "Yifan Bao", "Raad Khraishi", "Yihao Ang", "Anthony K. H. Tung", "Lukasz Szpruch", "Hao Ni"], "year": 2026, "published": "2026-08-18T15:55:32Z", "updated": "2026-08-18T15:55:32Z", "arxiv_id": "http://arxiv.org/abs/2608.17933v1", "url": "https://arxiv.org/abs/2608.17933v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CE"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Experiments across four benchmark datasets demonstrate that EvoTS-Agent consistently outperforms existing LLM-based agents while maintaining a 100\\% execution success rate across all evaluated backbone LLMs."], "abstract": "Financial time series exhibit non-stationary and heterogeneous statistical properties, making change-point detection challenging because no single unsupervised algorithm performs consistently across assets and market regimes. Conventional workflows consequently depend heavily on expert-driven model selection, feature design, and hyperparameter tuning, limiting their scalability and adaptability. We propose EvoTS-Agent, a validation-guided self-evolving LLM agent for autonomous financial time-series change-point detection. EvoTS-Agent first performs curated exploratory data analysis to characterize dataset properties and initialize candidate detection models. It then evolves executable experiment trajectories through three complementary operators: \\textit{Revision} exploits the current best solution, \\textit{Alternative Strategy} explores fundamentally different modeling directions when progress stagnates, and \\textit{Recombination} synthesizes complementary evidence from high-performing trajectories. Validation feedback guides trajectory evolution throughout the search, enabling the agent to adapt its detection pipeline to the statistical characteristics of each dataset while preserving reliable optimization. Experiments across four benchmark datasets demonstrate that EvoTS-Agent consistently outperforms existing LLM-based agents while maintaining a 100\\% execution success rate across all evaluated backbone LLMs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-c999a97c6c04ebf9", "title": "A Theoretical Framework for Parallel Lifelong MAPF Using Group Decentralized Planning", "authors": ["Alex DeWeese", "Jiaoyang Li", "Guannan Qu"], "year": 2026, "published": "2026-08-18T15:48:18Z", "updated": "2026-08-18T15:48:18Z", "arxiv_id": "http://arxiv.org/abs/2608.17928v1", "url": "https://arxiv.org/abs/2608.17928v1", "primary_category": "cs.MA", "categories": ["cs.MA", "cs.AI", "cs.RO"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Deep Research"], "benchmarks": [], "models": ["ying"], "key_results": ["We show that both RHCR and GD-RHCR achieve similar exponentially close to optimal guarantees, establishing a theoretical duality between the time based restrictions performed by vanilla RHCR and the additional space based partitioning performed by GD-RHCR."], "abstract": "In the Lifelong Multi-Agent Path Finding (L-MAPF) problem, agents must repeatedly move from one destination to another while avoiding obstacles and inter-agent collisions. Widely regarded as one of the highest-performing solutions to this problem is the Rolling-Horizon Collision Resolution (RHCR) framework. However, commensurate with its quality solutions, it incurs a computational cost that limits its applicability to even modest agent counts. In this paper, leveraging theoretical methods from the Locally Interdependent Multi-Agent MDP literature, we first theoretically prove the near-optimality of RHCR in a discounted MDP formulation of the L-MAPF problem. Then, we leverage these results to naturally motivate an extended framework called Group Decentralized RHCR (GD-RHCR) which incorporates a group decentralized structure that partitions agents based on a transitive communication scheme and plans for each partition of agents in parallel. We show that both RHCR and GD-RHCR achieve similar exponentially close to optimal guarantees, establishing a theoretical duality between the time based restrictions performed by vanilla RHCR and the additional space based partitioning performed by GD-RHCR. Lastly, we show that across varying maps, GD-RHCR is able to attain high throughput that scales into higher agent counts while maintaining a significantly lower per plan cost.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-c5e28c540d3749ce", "title": "AutoResearch: Insight In, Hallucination Out", "authors": ["Yiming Ren", "Xiang Liu", "Qumeng Sun", "Xiao Zhang", "Jiahao Li", "Haoyang Zhang", "Junjie Wang"], "year": 2026, "published": "2026-08-18T15:38:26Z", "updated": "2026-08-18T15:38:26Z", "arxiv_id": "http://arxiv.org/abs/2608.17906v1", "url": "https://arxiv.org/abs/2608.17906v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.MA"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Interpretability", "Deep Research"], "benchmarks": [], "models": [], "key_results": ["For example, on RSICD benchmark, an AutoResearch-generated idea improves mean Recall from 32.84 to 34.69, while recording only 5 audit-confirmed issue events compared with 11-27 for other autonomous research systems."], "abstract": "Autonomous research systems are increasingly capable of executing long research workflows, yet automation alone does not ensure that the resulting process remains scientifically grounded. We introduce AutoResearch, a two-stage system that connects Idea Generation with Idea Execution to address both how research ideas are formed and how they are reliably established through experimentation. In Idea Generation, AutoResearch continuously integrates emerging research signals with accumulated domain knowledge, identifies transferable mechanistic insights, and uses multi-model generation and cross-review to produce grounded, testable research plans. In Idea Execution, coordinated agents decompose these plans into experiments, iteratively implement and diagnose them, and employ independent evidence-based review before accepting research conclusions. Across representative settings in cross-modal retrieval, systems optimization, and benchmark-driven machine learning, AutoResearch turns generated ideas into measurable progress, detects and corrects unreliable experimental results, and makes evidence-conditioned decisions to continue, revise, or terminate research directions. For example, on RSICD benchmark, an AutoResearch-generated idea improves mean Recall from 32.84 to 34.69, while recording only 5 audit-confirmed issue events compared with 11-27 for other autonomous research systems. These results demonstrate a research process in which meaningful insight is grounded before experimentation and conclusions are grounded before acceptance: Insight In, Hallucination Out.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-213757132972aadb", "title": "AdaLens: Interactive Storyline for Monitoring and Steering Long-Running Agentic Data Analysis", "authors": ["Yangtian Liu", "Yan Miao", "Shuhan Liu", "Yunfan Zhou", "Dae Hyun Kim", "Di Weng", "Yingcai Wu"], "year": 2026, "published": "2026-08-18T14:34:45Z", "updated": "2026-08-18T14:34:45Z", "arxiv_id": "http://arxiv.org/abs/2608.17834v1", "url": "https://arxiv.org/abs/2608.17834v1", "primary_category": "cs.HC", "categories": ["cs.HC", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["Existing interactive approaches improve process visibility and open intervention points, but they remain largely designed for discrete, turn-by-turn exchanges rather than the parallel branches and evolving decision structures of long-running agentic analysis."], "abstract": "Large language models are pushing data science toward increasingly autonomous and agentic workflows, with recent systems already supporting multi-step and long-running analyses. As these workflows become more autonomous, conventional interfaces no longer provide adequate support for two critical requirements: observability for understanding an agent's evolving reasoning and evidence, and steerability for redirecting low-value directions or deepening promising ones during execution. Existing interactive approaches improve process visibility and open intervention points, but they remain largely designed for discrete, turn-by-turn exchanges rather than the parallel branches and evolving decision structures of long-running agentic analysis. We study this need as interactive oversight in long-running agentic data analysis and present AdaLens, an interactive system for monitoring and steering ongoing runs. AdaLens combines a storyline-based representation that unifies analytical plans, execution progress, intermediate findings, and data-column involvement with steering interactions grounded in these analytical elements for directional guidance and execution control. We evaluate AdaLens through two case studies and a user study, examining how it supports analysts in monitoring and steering long-running agentic data analysis.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-227bce157c620e3c", "title": "StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows", "authors": ["Liya Zhu", "Xin Ma", "Tao Liu", "Haodong Wang", "Ge Zhang", "Jingzhe Ding", "Qingshui Gu", "Yongjie Zhong", "Jinxiang Meng", "Yuan Gao", "Yunqiu Zhou", "Hao Zhu", "Jifeng He", "Yongzhi Liao", "Xinyi Zhang", "Chaoxin Li", "Yi Zhu", "Xi Lin", "Duju Zeng", "Xiang Gao", "Wen Zhang", "Yunyang Wang", "Duo Wang", "Huan Zhou", "Zuo Wang", "Jin Chen", "Kaiyuan Zhang", "Chuqian Yu", "Tianhao Yu", "Longxiang Liu", "Jianbo Xue", "Huimin Che", "Jiahao Wang", "Yujia Qin", "Jiaheng Liu", "Shen Yan", "Xiaolong Chang", "Wenhao Huang"], "year": 2026, "published": "2026-08-18T14:01:32Z", "updated": "2026-08-18T14:01:32Z", "arxiv_id": "http://arxiv.org/abs/2608.17800v1", "url": "https://arxiv.org/abs/2608.17800v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks."], "abstract": "Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing benchmarks largely rely on researcher-selected tasks, leaving uncertain whether such progress extends to the work that real-world users actually demand from AI systems. We introduce \\textbf{StartupBench}, an E2E agent benchmark grounded in market-validated AI startup products. Rather than defining tasks from pre-defined assumptions about useful agent capabilities, we systematically study AI products with demonstrated adoption, together with their product workflows and users, to identify real-world tasks for which AI has established practical demand across diverse professional domains. We translate these workflows into complete deliverable-oriented tasks and evaluate them with fine-grained rubrics capturing their complex requirements. Across representative models evaluated under a unified agent harness, even the strongest model successfully completes only approximately 30\\% of StartupBench, despite making substantial partial progress on many tasks. Further analysis identifies aspects like complex instruction following and domain-specific expertise as major sources of failure. Our results reveal that many market-validated workflows remain beyond the reliable capabilities of current general-purpose agents, establishing StartupBench as an empirical measure of progress toward E2E completions of real-world user tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-6f7e574a2593f865", "title": "TraceSQL: Traceable Answerability Estimation for Reference-Free Text-to-SQL Verification", "authors": ["Neelesh Kumar Shukla", "Debasmita Panda", "Srutanik Bhaduri", "Aditya Banerjee", "Viji Krishnamurthy"], "year": 2026, "published": "2026-08-18T13:56:29Z", "updated": "2026-08-18T13:56:29Z", "arxiv_id": "http://arxiv.org/abs/2608.17795v1", "url": "https://arxiv.org/abs/2608.17795v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["RLVR / GRPO", "Verifier / PRM", "Preference Optimization", "Interpretability"], "benchmarks": [], "models": [], "key_results": ["On BIRD development databases, TraceSQL achieves 66.47% F1 and 64.48% ROC-AUC, compared with 61.87% F1 and 58.26% ROC-AUC for the GradeSQL-7B ORM baseline on the same generated-SQL evaluation."], "abstract": "Text-to-SQL systems are commonly evaluated using ground-truth SQL queries or reference execution results, but such supervision is unavailable at inference time in real-world deployments. This creates a critical verification problem: given only a user question, database context, and generated SQL, can a system estimate whether the generated query is likely to correctly answer the question? Recent approaches use LLMs as judge or specialized agents to inspect generated SQL, but their decisions can be difficult to trace. Outcome Reward Models (ORMs) address this by learning from execution-labeled candidate SQLs and assigning correctness scores to unseen queries, yet they still provide limited visibility into the signals behind each verification. To address this limitation, we propose TraceSQL, a lightweight and traceable verification model built on explicit diagnostic features. TraceSQL combines 67 features capturing question ambiguity, question requirements, question-schema-SQL consistency, SQL structure, and intent alignment. These signals remain available for examining which factors influence each prediction and for tracing decisions back to diagnostic evidence. On BIRD development databases, TraceSQL achieves 66.47% F1 and 64.48% ROC-AUC, compared with 61.87% F1 and 58.26% ROC-AUC for the GradeSQL-7B ORM baseline on the same generated-SQL evaluation. Feature attribution further shows that the model relies on both semantic grounding and deterministic SQL-structure signals. These results show that SQL verification can be performed with a lightweight learned model while retaining feature-level evidence for inspecting and diagnosing its predictions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-e06d0d999926f345", "title": "D$^2$ACCI: A Dual-Loop Diagnostic Protocol for Evidence-Preserving Agent Memory", "authors": ["Xule Liu", "Yijun Liu", "Chao Li", "Shao Kun"], "year": 2026, "published": "2026-08-18T13:18:38Z", "updated": "2026-08-18T13:18:38Z", "arxiv_id": "http://arxiv.org/abs/2608.17756v1", "url": "https://arxiv.org/abs/2608.17756v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG"], "benchmarks": [], "models": ["yield"], "key_results": ["We instantiate the protocol in MemStack and evaluate on three public benchmarks, achieving 93.59% on LoCoMo, 90.93% on LongMemEval, and 57.20% on PersonaMem-V2.", "Five paired ablations show that supplement extraction, session-memory retrieval, and Forget Guard yield statistically significant gains (+1.9 to +3.7pp, all p $\\le$ .003).", "A diagnostic audit shows enriched traces substantially improve root-cause agreement over result-only relabeling."], "abstract": "Memory is a key capability of LLM agents. Persistent memory extends this across sessions---enabling recall, revision, and personalization. Yet its multi-stage pipeline (ingestion, retrieval, filtering, generation) makes failures difficult to localize: end-to-end evaluation reveals that an error occurred, but not which stage caused it. Existing evaluations often report aggregate performance without paired statistical comparisons, slice-level non-regression checks, or stage-level diagnostic traces. We propose D$^2$ACCI (Diagnostic-Driven Artifact-based Closed-loop Controlled Iteration), a dual-loop protocol whose outer diagnostic gate promotes, feature-flags, or rejects memory interventions based on paired evidence, protected-slice monitoring, and trace-level localizability. We further introduce DCR, a graded observability metric that measures whether failures remain localizable, and D$^2$ACCI-Eval, a reusable artifact for gate replay. We instantiate the protocol in MemStack and evaluate on three public benchmarks, achieving 93.59% on LoCoMo, 90.93% on LongMemEval, and 57.20% on PersonaMem-V2. Five paired ablations show that supplement extraction, session-memory retrieval, and Forget Guard yield statistically significant gains (+1.9 to +3.7pp, all p $\\le$ .003). In contrast, BM25/RRF is retained as a monitored feature flag---a distinction invisible to aggregate-only evaluation. A diagnostic audit shows enriched traces substantially improve root-cause agreement over result-only relabeling. Diagnostic artifacts reach 98--100% DCR@3 versus 0% for results-only logs. These results establish that robust memory-system iteration demands traceable, statistically grounded, and regression-aware evidence---exactly the gap D$^2$ACCI fills.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-bcc9eec4c73a1174", "title": "Optimize Your Sampling: Tuned Diffusion Sampling with Bayesian Optimization", "authors": ["Travis Zhang", "Christian Belardi", "Justin Lovelace", "Jin Peng Zhou", "Saebyeol Shin", "Carla P. Gomes", "Kilian Q. Weinberger"], "year": 2026, "published": "2026-08-18T17:33:00Z", "updated": "2026-08-18T17:33:00Z", "arxiv_id": "http://arxiv.org/abs/2608.18040v1", "url": "https://arxiv.org/abs/2608.18040v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Video Generation"], "benchmarks": [], "models": ["phisticated"], "key_results": ["OYS outperforms both the default schedules and those of Align Your Steps on text-to-image generation, and improves over the default schedules on inpainting and other image tasks, in both quantitative and human evaluations.", "OYS requires no additional training, is applicable even to distilled models, and improves both simple and sophisticated samplers such as Euler and DPM-Solver++."], "abstract": "Sampling from a diffusion model typically requires many forward passes through a large neural network, making generation computationally expensive. While much work has focused on efficient solvers and samplers, comparatively little attention has been paid to selecting the sampling timesteps themselves. A recent line of work optimizes theoretically derived surrogates for sample quality rather than the quality metric itself. We propose Optimizing Your Sampling (OYS), which instead treats timestep selection as a black-box optimization problem, optimizing the target metric directly with Bayesian optimization. OYS outperforms both the default schedules and those of Align Your Steps on text-to-image generation, and improves over the default schedules on inpainting and other image tasks, in both quantitative and human evaluations. OYS requires no additional training, is applicable even to distilled models, and improves both simple and sophisticated samplers such as Euler and DPM-Solver++. A 5-step OYS schedule retains 89%-94% of the quality of a 50-step schedule while reducing inference cost by 10x.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-c3cf265624a42b24", "title": "Dynamic Compression in Recurrent Networks", "authors": ["Jyothish Pari", "Ryan Bahlous-Boldi", "Pulkit Agrawal"], "year": 2026, "published": "2026-08-18T15:29:18Z", "updated": "2026-08-18T15:29:18Z", "arxiv_id": "http://arxiv.org/abs/2608.17896v1", "url": "https://arxiv.org/abs/2608.17896v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Long Context", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["We find that dynamic compression substantially reduces the recurrent state required for accurate reuse and scales more favorably as the number of stored functions grows."], "abstract": "Recurrent models process long contexts efficiently by compressing their history into a fixed-size state, but modern architectures typically do so in a single causal pass over the sequence. Each input must therefore be compressed before the model knows how it will later be used, forcing a limited state to compromise across possible future demands. We introduce dynamic compression, which allows a recurrent model to selectively revisit past tokens and revise its fixed-size state through additional recurrent updates. The model need not preserve every part of the history at uniformly high fidelity in its recurrent state, because lower-fidelity information can be revisited from the retained raw sequence when it becomes relevant. We study this in a controlled setting where the model first learns multiple functions in-context and, later in the same sequence, encounters a series of few-shot tasks that each require it to identify and reuse one of those functions. A single-pass model must preserve every function at sufficient fidelity for any future task, whereas selective re-scanning allows the model to revisit and refine only the function currently needed. We find that dynamic compression substantially reduces the recurrent state required for accurate reuse and scales more favorably as the number of stored functions grows. These results demonstrate a computation--memory tradeoff in which recurrent models can spend more computation revisiting their history to make more effective use of a fixed-size state.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-8954bac78a7cca47", "title": "Bounded-State Restoration: Decoupling Local Restore Capacity from External LLM State", "authors": ["Zixuan Li"], "year": 2026, "published": "2026-08-18T14:26:15Z", "updated": "2026-08-18T14:26:15Z", "arxiv_id": "http://arxiv.org/abs/2608.17826v1", "url": "https://arxiv.org/abs/2608.17826v1", "primary_category": "cs.DC", "categories": ["cs.DC"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache", "Long Context", "Memory / RAG"], "benchmarks": [], "models": ["DeepSeek-V4-Flash"], "key_results": ["A matched SSD optimization reduces 512K restore TTFT from 43.1 to 17.6 seconds without changing RWS."], "abstract": "Hierarchical KV-cache systems can retain long-context LLM execution state beyond GPU memory, but retention capacity does not determine the local memory required to make that state executable again. We isolate this second resource as the restoration working set (RWS): the peak local staging state whose lifetimes overlap during restoration. In the pinned upstream LMCache whole-plan path, measured full-reuse points for 1.956, 7.823, and 15.646 GiB/rank states first succeed at 2, 8, and 16 GiB L1 rungs, with successful L1 peaks of 1.956, 7.824, and 15.648 GiB/rank.\n  We introduce Bounded-State Restoration (BSR), which separates complete discovery from local residency. BSR probes the complete reusable prefix without materializing the whole hit in L1, then installs confirmed state through a reusable window of at most $W$ chunks. Under bounded auxiliary state, peak restoration capacity is $O(W)$ while total transfer and installation work remains $Θ(|S|)$. Because reusable state spans heterogeneous allocator groups and tensor-parallel ranks, BSR uses a request-level commit rule: partial installation is never exposed as a valid reusable prefix; failures invalidate the advertised prefix and fall back to a lower valid tier or deterministic recomputation.\n  On DeepSeek-V4-Flash with TP=2 across two DGX Spark nodes, a clean no-resume sweep grows external state from 1.956 to 31.277 GiB/rank while measured L1 RWS remains exactly 500.75 MiB/rank at $W=32$, a 63.959x largest-state external-to-live-staging ratio. A second fresh 524K-token run repeats the largest-state acceptance result. Evaluated tier and rank-asymmetric failures expose either complete reuse or zero external reuse before fallback. A matched SSD optimization reduces 512K restore TTFT from 43.1 to 17.6 seconds without changing RWS.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-b7f027e5e87fe637", "title": "Thinking in a Low-Resource Language: What SFT Builds, What RL Fixes, What Accuracy Cannot See", "authors": ["Ayoub Kirouane", "Christos Petrocheilos"], "year": 2026, "published": "2026-08-18T13:09:03Z", "updated": "2026-08-18T13:09:03Z", "arxiv_id": "http://arxiv.org/abs/2608.17744v1", "url": "https://arxiv.org/abs/2608.17744v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG", "cs.RO", "stat.ML"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "Chain-of-Thought", "MoE", "Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Reinforcement learning with verifiable rewards, pre-registered before training, fixes the first two outright (fallback 24% to 2.5%, leak 3.5% to 0.0%, both against a flat random-reward control) and moves the third (+9.1pp), while the Greek reasoning habit survives an accuracy-only gradient untouched."], "abstract": "Take three frontier mixture-of-experts models (Alibaba, OpenAI, NVIDIA; 3.6-4.0B active parameters each) and fine-tune them to reason in a low-resource language. On accuracy benchmarks almost nothing happens, and the benchmark itself is noise at this scale: changing only the random seed moves the score by 7.7 points, more than every data and recipe effect we measured. That null is our first result. The real changes live where accuracy cannot see. Base models never think in Greek: 0 of 1,000 reasoning traces, even when the question is Greek, so the model answers correctly while reasoning in a form its user cannot read, audit, or correct. After supervised fine-tuning (SFT), every released checkpoint reasons in the language of the question on ~98% of items, one family at 3x fewer tokens, with judged grammaticality improving on all four models and general ability within a few points of each base: nothing was forgotten, and fluency was gained. We propose six behavioural dimensions that make such changes measurable, each gated to reject any metric that correlates with output length, and we report how our own instruments lied: six failures, each caught by a control. What SFT cannot do is fix its own defects: a quarter of answers skip the requested format, answers leak into the reasoning channel, and an explicit \"think in English\" is obeyed under half the time. Reinforcement learning with verifiable rewards, pre-registered before training, fixes the first two outright (fallback 24% to 2.5%, leak 3.5% to 0.0%, both against a flat random-reward control) and moves the third (+9.1pp), while the Greek reasoning habit survives an accuracy-only gradient untouched. We release five checkpoints. The instruments, the controls and the pre-registration travel to any low-resource language; Greek is the case that let us measure them.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-24812cd00150a83f", "title": "Mixture-of-Expert Blocks Contain Strong Hallucination Detection Signals", "authors": ["Joao Fonseca", "Rodrigo Rodrigues", "Paolo Romano"], "year": 2026, "published": "2026-08-18T12:00:20Z", "updated": "2026-08-18T12:00:20Z", "arxiv_id": "http://arxiv.org/abs/2608.17687v1", "url": "https://arxiv.org/abs/2608.17687v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["MoE"], "benchmarks": [], "models": [], "key_results": ["Our results show that InnerExpert outperforms existing methods across five datasets and two MoE architectures, achieving up to 0.91 answer-level and 0.76 token-level AUROC, while requiring only a single forward pass."], "abstract": "Despite their widespread use, Large Language Models (LLMs) remain limited by a fundamental problem: the generation of plausible but false content, known as hallucinations. Most existing detection methods operate at the answer or sentence level, yet per-token detection is essential for localizing hallucinated spans and enabling fine-grained interventions. In this paper, we explore the use of the Mixture-of-Experts (MoE) paradigm to address this gap. In MoE architectures, a single forward pass activates a sparse subset of experts (i.e., distinct feedforward networks per layer) via a routing mechanism, producing internal signals (e.g., router entropy, expert disagreement, and expert usage patterns) that are unavailable in dense architectures and have not been previously exploited for hallucination detection. To this end, we introduce InnerExpert, the first method to leverage these MoE-specific signals for per-token hallucination detection. InnerExpert combines routing-level and standard transformer signals into compact per-token feature vectors, classified by a lightweight detector trained on labels produced by an LLM-as-a-judge pipeline, which enables continuous model updates without manual annotation. Our results show that InnerExpert outperforms existing methods across five datasets and two MoE architectures, achieving up to 0.91 answer-level and 0.76 token-level AUROC, while requiring only a single forward pass.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-9f3e879c99dbbf23", "title": "MoNe: Modular Neural Memory for Efficient Long Context Inference", "authors": ["Wonguk Cho", "Kyubyung Chae", "Tribhuvanesh Orekondy", "Sunghyun Park", "Hyoungwoo Park", "Jeongho Kim", "Arash Behboodi", "Kyuwoong Hwang", "Sungrack Yun"], "year": 2026, "published": "2026-08-18T10:28:23Z", "updated": "2026-08-18T10:28:23Z", "arxiv_id": "http://arxiv.org/abs/2608.17616v1", "url": "https://arxiv.org/abs/2608.17616v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Test-time Scaling", "Long Context", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["This two-phase design decouples inference cost from context length, achieving $O(N)$ preprocessing and $O(1)$ query cost with peak GPU memory that does not grow with $N$.", "At 128K tokens, MoNe reduces both compute and peak GPU memory by approximately 80% compared to ICL with only 6.4% parameter overhead.", "MoNe generalizes to context lengths far beyond the backbone's native window, achieving strong performance on needle-in-a-haystack and word extraction benchmarks from RULER, where ICL degrades sharply."], "abstract": "We present MoNe, a lightweight modular neural memory that attaches to any frozen pretrained Transformer to enable long-context inference without retraining. MoNe reads context in fixed-size segments via test-time learning of fast-weight neural memory networks with layer-localized gradient updates; at inference, the memory generates keys and values from the query tokens alone, with no context tokens re-read. This two-phase design decouples inference cost from context length, achieving $O(N)$ preprocessing and $O(1)$ query cost with peak GPU memory that does not grow with $N$. At 128K tokens, MoNe reduces both compute and peak GPU memory by approximately 80% compared to ICL with only 6.4% parameter overhead. MoNe generalizes to context lengths far beyond the backbone's native window, achieving strong performance on needle-in-a-haystack and word extraction benchmarks from RULER, where ICL degrades sharply.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-366d1558cf7a4f8e", "title": "ArborMem: Navigating Interaction States with Memory Forests", "authors": ["Zongwei Lv", "Yuemeng Xu", "Yilun Yao", "Siyi Ding", "Xinyu Tan", "Yaoming Li", "Guangxiang Zhao", "Weihong Lin", "Lin Sun", "Xiangzheng Zhang", "Tong Yang"], "year": 2026, "published": "2026-08-18T08:54:55Z", "updated": "2026-08-18T08:54:55Z", "arxiv_id": "http://arxiv.org/abs/2608.17534v1", "url": "https://arxiv.org/abs/2608.17534v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Long Context", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Existing methods improve access to conversational history through long-context processing, selective retrieval, and structured memory organization.", "Experiments on LongMemEval, LoCoMo, BEAM 100K, and BranchMemEval show that ArborMem outperforms the strongest baselines by 3.36 to 10.31 percentage points on the three established benchmarks and by 5.0 points on BranchMemEval."], "abstract": "Large language models increasingly serve as persistent conversational assistants, requiring memory that preserves relevant experience and maintains continuity across interactions. Existing methods improve access to conversational history through long-context processing, selective retrieval, and structured memory organization. However, most systems treat memory access as retrieving relevant past information without first determining which prior interaction state the current turn resumes. This limitation becomes particularly important when conversations interleave multiple tasks, people, and plans that may be interrupted and later revisited. We introduce ArborMem, an online memory framework that represents a long-running conversation as a navigable forest of interaction states. Each branch preserves a locally coherent trajectory, while the forest maintains multiple trajectories that may later be resumed. For each new input, ArborMem localizes the relevant state, restores its branch-local context, and augments it with reusable evidence retrieved across branches, preserving interaction continuity without conflating semantically related but structurally distinct trajectories. Existing long-term memory benchmarks cover diverse memory and reasoning capabilities but do not explicitly isolate branch-structured challenges. We therefore introduce BranchMemEval, a controlled diagnostic benchmark for interleaved and resumable interaction trajectories. Experiments on LongMemEval, LoCoMo, BEAM 100K, and BranchMemEval show that ArborMem outperforms the strongest baselines by 3.36 to 10.31 percentage points on the three established benchmarks and by 5.0 points on BranchMemEval. Its advantage grows under constrained read budgets, while complete memory queries remain below half a second.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-9493f9dda6626380", "title": "Integrating Novelty and Surprise for Experience Prioritization and Exploration in Image-Based Reinforcement Learning", "authors": ["Hoda Yamani", "Henry Williams", "Bruce A. MacDonald"], "year": 2026, "published": "2026-08-18T05:07:58Z", "updated": "2026-08-18T05:07:58Z", "arxiv_id": "http://arxiv.org/abs/2608.17373v1", "url": "https://arxiv.org/abs/2608.17373v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["We further extend this with NSPER+R, integrating these signals as intrinsic rewards to jointly improve replay quality and exploration.", "Experiments on DeepMind Control Suite tasks show that NSPER and NSPER+R improve training efficiency and convergence speed compared to existing methods in image-based RL."], "abstract": "Sample efficiency is a central challenge in reinforcement learning (RL), particularly in image-based domains where agents must learn from high-dimensional visual inputs. Traditional sampling often relies on random or suboptimal experience selection, leading to redundant updates and slow learning. Improving efficiency requires mechanisms that prioritize informative experiences while also encouraging effective exploration. Prioritized Experience Replay (PER) addresses part of this challenge by reusing high-value transitions, while intrinsic rewards promote the exploration of novel or uncertain states. However, their integration has not been extensively studied. This paper introduces Novelty and Surprise Prioritized Experience Replay (NSPER), which uses novelty to capture underrepresented states and surprise to expose gaps in the agent's understanding of the environment. We further extend this with NSPER+R, integrating these signals as intrinsic rewards to jointly improve replay quality and exploration. Experiments on DeepMind Control Suite tasks show that NSPER and NSPER+R improve training efficiency and convergence speed compared to existing methods in image-based RL.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-a5530e500a5ba464", "title": "MoFE: A Novel Mixture-of-Experts Framework with Fourier Neural Operators for Cryptocurrency Forecasting", "authors": ["Bowen Liu", "Mingming Sun"], "year": 2026, "published": "2026-08-18T04:00:16Z", "updated": "2026-08-18T04:00:16Z", "arxiv_id": "http://arxiv.org/abs/2608.17342v1", "url": "https://arxiv.org/abs/2608.17342v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["MoE", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["Extensive experiments on Bitcoin datasets spanning January 2020 to December 2025 demonstrate that MoFE achieves state-of-the-art (SOTA) performance in both T+1 and T+5 forecasting horizons.", "In high-fidelity simulated trading environments, these predictive gains transfer into significant excess returns and robust risk-adjusted performance, characterized by a high Sharpe ratio."], "abstract": "Forecasting cryptocurrency prices remains a formidable challenge due to inherent non-stationarity, abrupt regime shifts, and multi-scale stochastic dependencies. Conventional deep learning models often struggle to capture complex underlying dynamics, frequently resulting in persistent phase-lagged predictions. To address these limitations, we propose MoFE, a novel deep learning framework that integrates Fourier Neural Operators (FNOs) within a Mixture-of-Experts (MoE) architecture. Rooted in the theoretical framework of stochastic differential equations, MoFE conceptualizes cryptocurrency volatility as a superposition of multi-frequency components, which includes user network based fundamental growth, mining costs and halving mechanism caused seasonal volatility, and market sentiment-induced chaos. Specifically, specialized adaptive FNO (AFNO) and Convolution dual-domain experts learn continuous function-to-function mappings to encapsulate global spectral trends, cyclical adjustments and microstructures, while a dynamic gating based MoE mechanism enables adaptive strategy switching across diverse market regimes. Extensive experiments on Bitcoin datasets spanning January 2020 to December 2025 demonstrate that MoFE achieves state-of-the-art (SOTA) performance in both T+1 and T+5 forecasting horizons. Notably, the model effectively mitigates the phase-lag effect, delivering superior Directional Accuracy (DA) and Information Coefficient (IC). In high-fidelity simulated trading environments, these predictive gains transfer into significant excess returns and robust risk-adjusted performance, characterized by a high Sharpe ratio.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-10c0126896e6737c", "title": "Abra: Scaling Diffusion Image Training", "authors": ["Kyle Chickering", "Wei-An Lin", "Swayam Bhanded", "Dan Saunders", "Akshat Tripathi", "Jiaming Song", "Shyamal Buch", "Xinchen Yan"], "year": 2026, "published": "2026-08-18T02:36:25Z", "updated": "2026-08-18T02:36:25Z", "arxiv_id": "http://arxiv.org/abs/2608.17286v1", "url": "https://arxiv.org/abs/2608.17286v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Test-time Scaling", "Video Generation"], "benchmarks": [], "models": [], "key_results": ["Compute-optimal scaling laws guide the training of frontier language models yet remain largely unexplored for visual generation."], "abstract": "Compute-optimal scaling laws guide the training of frontier language models yet remain largely unexplored for visual generation. We present a systematic scaling law study for text-to-image diffusion models using Abra, a controlled family of flow-matching transformers trained across three orders of magnitude worth of compute ($10^{19}$ to $10^{22}$ FLOPs), reaching significantly larger compute budgets than previous works. We demonstrate that diffusion models scale just as predictably as language models but require far more data to train optimally: compute optimality occurs at approximately $200$ image tokens per parameter, ten times the Chinchilla compute-optimal prescription for LLMs. We show that unlike language models, diffusion models are robust to overtraining and that practitioners should err on the side of more data rather than a larger model. Finally, we show that this predictability extends beyond training loss to generative quality metrics, optimal CFG settings, representation quality, and even the shape of the training curves, which collapse onto a universal form.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-4dc5ce2927df9735", "title": "Teach and Grow: An Agent-Centered Architecture for General Robot Learning", "authors": ["Chang Nie", "Zhe Liu", "Hesheng Wang"], "year": 2026, "published": "2026-08-17T23:45:21Z", "updated": "2026-08-17T23:45:21Z", "arxiv_id": "http://arxiv.org/abs/2608.17209v1", "url": "https://arxiv.org/abs/2608.17209v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI", "cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Memory / RAG", "VLM"], "benchmarks": [], "models": [], "key_results": ["Our LIBERO evaluation attains state-of-the-art performance; controlled studies expose skill induction, persistent reuse, and agent-directed adaptation."], "abstract": "End-to-end vision-language-action (VLA) and world-action models offer an elegant route to general-purpose robotics, but their reliability is bounded by validated physical coverage. When an unfamiliar object, sensor, embodiment, or contact falls outside that coverage and no validated fallback exists, correcting the failure requires new robot data, a policy update, and regression testing. This recurring burden is the retraining tax. Unlike text, embodied data must often be created by operating machines. We present Teach-and-Grow Learning (TGL), an agent-centered architecture for general robot learning. In its general form, a multimodal agent turns a few successful demonstrations into reusable Skill Blocks: closed-loop behaviors for meaningful subgoals. In a new scene, the agent grounds and composes these blocks, selects learned or geometric tools, observes the physical outcome, and revises the route when execution departs from intent. A Skill Library stores executable behavior, while structured Experience Memory carries forward success, failure, and repair. New tasks are acquired without task-specific policy retraining. Our LIBERO evaluation attains state-of-the-art performance; controlled studies expose skill induction, persistent reuse, and agent-directed adaptation. Finally, we propose the Teach-and-Grow scaling-law hypothesis: if X denotes effective reusable experience, future-task error and teaching demand should approach irreducible floors as power laws in X. The architecture therefore treats deployment as a period of continued learning, in which one task can make the next easier.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-6778180fcf0a4de0", "title": "Lymphocyte Mimicry Correction via Region-Level Tissue Reasoning and Unbalanced Optimal Transport", "authors": ["Xiang Li", "Yuqi Wang", "Casey C. Heirman", "Jihye Heo", "Kyle J. Lafata"], "year": 2026, "published": "2026-08-17T21:33:55Z", "updated": "2026-08-17T21:33:55Z", "arxiv_id": "http://arxiv.org/abs/2608.17151v1", "url": "https://arxiv.org/abs/2608.17151v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["On the independent TCGA-BRCA cohort, Loki-OT achieved lower patient-level MAE than the fully supervised in-domain PanopTILs classifier and improved F1 in epithelium-rich mimicry tissues, using 278 weak region-level MLLM estimates built on a general-domain cell foundation model."], "abstract": "Cell mimicry arises when different cell types appear morphologically similar. Human pathologists resolve this ambiguity using surrounding tissue context, whereas current vision models either lack contextual reasoning (cell foundation models) or cannot operate at the cell level (pathology MLLMs). We present Loki-OT, which propagates region-level tissue reasoning to individual cell predictions via Unbalanced Optimal Transport, using MLLM-derived density priors as soft guidance for ambiguous cell reassignment. Loki-OT is motivated by the observation that pretrained cell foundation model features already encode discriminative information, including tissue context, but standard cell-level supervision fails to use tissue context effectively. The resulting transport plan is distilled into a lightweight student MLP classifier that learns context-aware decision boundaries within the pretrained feature space. On the independent TCGA-BRCA cohort, Loki-OT achieved lower patient-level MAE than the fully supervised in-domain PanopTILs classifier and improved F1 in epithelium-rich mimicry tissues, using 278 weak region-level MLLM estimates built on a general-domain cell foundation model. Code: https://github.com/xiangli980/Lymphocyte_Mimicry_Correction_via_Loki_OT", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-e769f145e13464f5", "title": "Children, but not language models, show accelerating returns in word learning", "authors": ["Michael C. Frank"], "year": 2026, "published": "2026-08-17T20:55:30Z", "updated": "2026-08-17T20:55:30Z", "arxiv_id": "http://arxiv.org/abs/2608.17120v1", "url": "https://arxiv.org/abs/2608.17120v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": [], "benchmarks": [], "models": [], "key_results": ["Children learn hundreds of words over the first years of their lives, in a process that begins slowly but quickly picks up speed."], "abstract": "Children learn hundreds of words over the first years of their lives, in a process that begins slowly but quickly picks up speed. Prior models describe vocabulary growth as evidence accumulation over time. Here we show that the process is best characterized as accelerating accumulation: children learn more from each additional unit of linguistic experience than they did from the one before. In contrast to children, language models -- even those trained on child-directed speech -- do not accelerate. Instead, they show constant proportional returns on new data, consistent with scaling laws. Children learn using many orders of magnitude less training data than language models; their increasingly efficient use of their learning input is a candidate explanation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-1d61056cf5c500ee", "title": "J-Miner: Recovering Executable Decision Knowledge from Language-Model Classifiers", "authors": ["Yunfan Gao", "Xinyi Huang", "Tao Sheng", "Haorui Song", "Yun Xiong", "Haofen Wang"], "year": 2026, "published": "2026-08-17T19:09:38Z", "updated": "2026-08-17T19:09:38Z", "arxiv_id": "http://arxiv.org/abs/2608.17063v1", "url": "https://arxiv.org/abs/2608.17063v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation"], "benchmarks": [], "models": [], "key_results": ["Across multiple classification tasks, J-Miner rules reproduce up to 98.3\\% of source-classifier decisions and achieve 6.0--29.5 percentage points higher behavioral fidelity than equally compact rules learned from input words."], "abstract": "Large language models can be fine-tuned into specialized classifiers that perform well across diverse text tasks and make complex judgments, but they typically expose only final labels, leaving the decision knowledge acquired through fine-tuning implicit within the model. We study how to mine this internal decision knowledge from a fine-tuned classifier and encode it in an executable representation that can be inspected, validated, and reused beyond the source classifier. We introduce J-Miner, which mines text-level named concepts by aggregating vocabulary-aligned internal signals across layers and token positions, and uses the classifier's own predictions to learn executable decision rules over them. This process distills local internal readouts into an explicit classifier-level knowledge representation. Across multiple classification tasks, J-Miner rules reproduce up to 98.3\\% of source-classifier decisions and achieve 6.0--29.5 percentage points higher behavioral fidelity than equally compact rules learned from input words. Further analysis shows that the named concepts reflect internal semantic evidence associated with task decisions, while the learned rules consolidate these distributed signals into inspectable decision structures. The resulting decision knowledge also transfers to lightweight standalone students: using about 1/24 as many parameters as the source classifiers, they reconstruct and execute the representation from raw text while retaining 99.8\\% of the source classifiers' mean task accuracy. These findings show that task-specific decision knowledge can be faithfully represented in an explicit, executable form and reused beyond the classifier in which it was learned.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-5e1fb3dd62375ebc", "title": "Primitive Representation Learning for Unsupervised Dynamic Contrast Enhanced MRI Reconstruction", "authors": ["Veronika Spieker", "Wenqi Huang", "Cemre Ariyurek", "Liam Timms", "Daniel Rueckert", "Onur Afacan", "Julia A. Schnabel", "Sila Kurugol"], "year": 2026, "published": "2026-08-18T17:48:22Z", "updated": "2026-08-18T17:48:22Z", "arxiv_id": "http://arxiv.org/abs/2608.18055v1", "url": "https://arxiv.org/abs/2608.18055v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV", "cs.LG", "eess.SP", "physics.med-ph"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["We show that this architecture achieves performance competitive with conventional reconstruction methods, both in reconstruction quality and in the accuracy of extracted aorta and kidney enhancement curves."], "abstract": "Reliable quantitative analysis of dynamic contrast-enhanced MRI requires high-quality spatiotemporal reconstructions at high undersampling rates. Scan-specific reconstructions using Gaussian and Gabor primitives have shown promising results without the need for large training datasets, but have not addressed the additional dimension of dynamic contrast. We propose a multi-dimensional, primitive based framework for dynamic contrast-enhanced MRI reconstruction that disentangles the underlying anatomy, the dynamic contrast enhancement, and residual motion into separate temporal basis functions, thereby enabling a geometrical interpretation of the representation. We show that this architecture achieves performance competitive with conventional reconstruction methods, both in reconstruction quality and in the accuracy of extracted aorta and kidney enhancement curves. The modular tier design extends naturally to additional dynamic factors and higher acceleration rates. Code available at https://github.com/compai-lab/ 2026-GaborDCE-spieker.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-4dbd18782dd97cc5", "title": "Where A Small Language Model Helps in Invoice Categorisation, Understood Through Embedding Geometry", "authors": ["Emma Ceccherini", "Daniel Lawson", "Anjulika Salhan"], "year": 2026, "published": "2026-08-18T17:28:01Z", "updated": "2026-08-18T17:28:01Z", "arxiv_id": "http://arxiv.org/abs/2608.18033v1", "url": "https://arxiv.org/abs/2608.18033v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Quant / Trading"], "benchmarks": [], "models": ["BERT"], "key_results": ["Whilst AI is increasingly being adopted across industries to automate tasks, including invoice categorisation, implementations built on in-house small language models (SLMs) can simultaneously reduce cost and improve data security, confidentiality, and interpretability.", "Combining these results with geometric analysis shows that pre-trained embedding geometry is associated with classification performance and reveals a counterintuitive finding that a structured input that would help a human reader does not improve the SLM performance."], "abstract": "Categorising invoices into the correct General Ledger (GL) code underpins financial reporting and tax compliance. This is a skilled accounting judgement rather than a routine task: the correct category depends subtly on the nature of the purchasing business, the vendor and the invoice text. Whilst AI is increasingly being adopted across industries to automate tasks, including invoice categorisation, implementations built on in-house small language models (SLMs) can simultaneously reduce cost and improve data security, confidentiality, and interpretability. We investigate this approach by first analysing the pre-trained embedding geometry of a small sentence transformer (SBERT) and classic SLM (DeBERTa). The sentence-embedding space of this financial corpus is globally anisotropic but composed of locally isotropic clusters, extending prior token-level findings to sentence embeddings in a financial setting, and these clusters are strongly correlated with the vendor identity. SBERT fine-tuned on a single GPU reaches 0.96 accuracy on invoice classification, above both a zero-shot LLM and a vendor identity baseline, increasing performance for smaller, challenging categories and new clients. For this important generalisation problem, SBERT reaches 0.9 F1 with roughly 100 client-specific invoices, showing that an in-house SLM implementation is promising. Combining these results with geometric analysis shows that pre-trained embedding geometry is associated with classification performance and reveals a counterintuitive finding that a structured input that would help a human reader does not improve the SLM performance.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-276b138c9970d33c", "title": "Judge, Retrieve, or Abstain: Uncertainty-Guarded LLM Judging with Provable Risk Guarantees", "authors": ["Sher Badshah", "Ali Emami", "Hassan Sajjad"], "year": 2026, "published": "2026-08-18T16:42:02Z", "updated": "2026-08-18T16:42:02Z", "arxiv_id": "http://arxiv.org/abs/2608.17994v1", "url": "https://arxiv.org/abs/2608.17994v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Memory / RAG"], "benchmarks": [], "models": ["ying"], "key_results": ["Across open-domain QA benchmarks and judges of varying scales, the framework maintains the target error rate while achieving substantially higher coverage than single-mode baselines."], "abstract": "Using LLMs as judges has become standard practice for evaluating model outputs at scale. This is particularly common for subjective, open-ended tasks such as assessing helpfulness or alignment, where no single reference answer exists. However, objective tasks introduce a distinct reliability challenge for reference-free LLM judging. In the absence of a reference answer, the judge evaluates factual correctness either through its parametric knowledge or through tool augmentation. Although the former enables efficient evaluation, the judge may hallucinate or lack sufficient evidence for its verdict. Conversely, tool augmentation can provide additional evidence but introduces extra computational cost and requires an appropriate mechanism to determine when and how that evidence should be used reliably. More importantly, neither approach alone provides formal control over the risk of accepted verdicts or guarantees their reliability at a specified level. We propose a risk-controlled framework that calibrates uncertainty thresholds on a held-out set so that the false discovery rate among accepted verdicts remains below a user-specified level~$α$ with high probability, using finite-sample Clopper--Pearson intervals. When the parametric mode is not sufficiently confident, the instance is routed to a retrieval-augmented mode, where the judge gathers web evidence and re-evaluates the instance under a second calibrated threshold. The finite-sample guarantee carries over to this two-threshold routing without additional assumptions. Across open-domain QA benchmarks and judges of varying scales, the framework maintains the target error rate while achieving substantially higher coverage than single-mode baselines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-360e842da13deb6e", "title": "A Residual Learning Approach for Unsteady Aerodynamic Load Prediction", "authors": ["Divya Sanghi", "Carlos E. S. Cesnik"], "year": 2026, "published": "2026-08-18T15:26:55Z", "updated": "2026-08-18T15:26:55Z", "arxiv_id": "http://arxiv.org/abs/2608.17894v1", "url": "https://arxiv.org/abs/2608.17894v1", "primary_category": "physics.flu-dyn", "categories": ["physics.flu-dyn", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["This paper investigates the feasibility of using residual learning to improve unsteady aerodynamic load prediction for aeroelastic applications."], "abstract": "This paper investigates the feasibility of using residual learning to improve unsteady aerodynamic load prediction for aeroelastic applications. The machine learning technique selected for the study is the long short-term memory (LSTM) neural network, which is used for its suitability for sequential data with aerodynamic memory effects. The approach is investigated for the NLR 7301 airfoil benchmark using high-fidelity CFD lift data for prescribed pitch and plunge motions in the transonic flow regime in the presence of shock motion. An analytical unsteady aerodynamic model based on the Wagner function is used as a physics-based baseline, and the neural network is trained to learn the difference between the CFD lift coefficient and the Wagner prediction. The residual model is compared with a direct neural-network model trained to predict the CFD lift coefficient. The comparison includes feature and normalization studies, external benchmark cases, and leave-one-out and leave-family-out generalization tests across a range of sinusoidal and non-sinusoidal motions. The residual model performs best when its inputs align with the Wagner formulation variables, generally giving lower error and more consistent performance across training runs, though the direct model remains more accurate for some high-frequency cases. The residual model also generalizes better in the leave-one-out and leave-family-out tests, with a smaller increase in error than the direct model when entire motion families are withheld from training. Overall, the results indicate that residual learning shows promise as a modular approach for augmenting classical low-order aerodynamic theories, especially when the physics baseline removes a structured part of the aerodynamic response and leaves a lower-variance correction for the neural network to learn.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-50ac0b12e4a8fed3", "title": "BayesPrompt: human readable prompts that make sense", "authors": ["Franky Kevin Nando Tezoh", "Ali Hussaini Umar", "Alessandro Laio", "Guido Sanguinetti", "Riccardo Rende"], "year": 2026, "published": "2026-08-18T14:59:04Z", "updated": "2026-08-18T14:59:04Z", "arxiv_id": "http://arxiv.org/abs/2608.17866v1", "url": "https://arxiv.org/abs/2608.17866v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["yield"], "key_results": ["We compare our approach with state of the art alternatives showing on a real data set a marked improvement over a range of metrics."], "abstract": "Reconstructing prompts that can elicit a desired answer or behaviour in an LLM is an open and important research topic. Optimisation methods which aim at minimising the perplexity of a given answer, however, consistently yield so-called pseudoprompts, unintelligible strings of tokens which can lack human interpretability. We argue that this is a consequence of the ill-posedness of the prompt optimisation task. By reframing the task as a Bayesian posterior inference over prompts, we propose an efficient algorithm to sample prompts which are both efficient (in terms of perplexity) and human readable. We compare our approach with state of the art alternatives showing on a real data set a marked improvement over a range of metrics.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-1b5e642162cd023e", "title": "Interpretable Humans, Alien LLMs: Expert Analysis of Latent Structures in Assessment Responses", "authors": ["Alona Strugatski", "Licol Zeinfeld", "Jason Cooper", "Shelley Rap", "Gil Schwarts", "Giora Alexandron"], "year": 2026, "published": "2026-08-18T14:08:25Z", "updated": "2026-08-18T14:08:25Z", "arxiv_id": "http://arxiv.org/abs/2608.17810v1", "url": "https://arxiv.org/abs/2608.17810v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.HC"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["The evaluation of large language models (LLMs) relies heavily on human-designed assessments, implicitly assuming that AI and humans employ similar underlying cognitive constructs."], "abstract": "The evaluation of large language models (LLMs) relies heavily on human-designed assessments, implicitly assuming that AI and humans employ similar underlying cognitive constructs. Challenging this assumption, we investigate whether the latent factors governing LLM performance carry the same substantive, human-interpretable meaning as the cognitive constructs governing human learners. Using responses from humans and six LLMs across quantitative reasoning and chemistry assessments, we conducted Exploratory Factor Analysis (EFA) separately for both groups. Subject-Matter Experts (SMEs) then blindly evaluated the resulting factor graphs to ascribe pedagogical meaning to the emerged constructs. SMEs successfully interpreted most of the human-derived factors. Conversely, they could not ascribe meaning to any LLM-derived factors in quantitative reasoning and interpreted only half of the LLM factors in chemistry. By combining data-driven EFA with blind expert interpretation, this framework shows that LLMs frequently operate on statistically opaque mechanisms distinct from human reasoning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-5047be4dc9e35fac", "title": "An Empirical Study of Reward Specification and Benchmark Reliability in GRPO-based LLM Unlearning", "authors": ["Rubén Balbastre", "Juan Manuel Orduña", "Mariano Pérez"], "year": 2026, "published": "2026-08-18T14:04:29Z", "updated": "2026-08-18T14:04:29Z", "arxiv_id": "http://arxiv.org/abs/2608.17804v1", "url": "https://arxiv.org/abs/2608.17804v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Practical LLM unlearning is usually evaluated through two objectives: suppress target-specific knowledge and preserve non-target utility."], "abstract": "Practical LLM unlearning is usually evaluated through two objectives: suppress target-specific knowledge and preserve non-target utility. In generative QA, this leaves a third behavior underspecified: when a target-adjacent prompt admits a broader answer without target-specific leakage, the model should answer at that level rather than leak, evade, or refuse. We study this specification problem in a controlled LoRA-GRPO RWKU setting, comparing four reward designs that span lexical suppression, anti-refusal shaping, rubric-based broad answering, and an explicit refusal contrast, with and without SFT warm-up. The experiments show that optimization success is not equivalent to behavioral unlearning: RWKU forget scores, held-out completion audits, terminal training-rollout audits, and training dynamics can point to different conclusions. We trace these disagreements to reward-hacking endpoints, policy-support limits in GRPO, benchmark probes that miss endpoint changes, and rewards that can select broad-topic answering with low semantic leakage during optimization.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-589bfac8d3bf4e81", "title": "Debate Training Reduces Reward Hacking in RLAIF", "authors": ["Zachary Kenton", "Lili Janzer", "Rory Greig", "Tian Huey Teh", "Kirill Tyshchuk", "Jonah Brown-Cohen", "Harri Edwards", "Senthooran Rajamanoharan", "Noah Y. Siegel", "Natasha Jaques", "Rohin Shah"], "year": 2026, "published": "2026-08-18T13:40:29Z", "updated": "2026-08-18T13:40:29Z", "arxiv_id": "http://arxiv.org/abs/2608.17776v1", "url": "https://arxiv.org/abs/2608.17776v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["RLVR / GRPO", "Preference Optimization", "Multi-Agent"], "benchmarks": [], "models": ["Gemini"], "key_results": ["We demonstrate that RL finetuning an LLM using debate, a two-player adversarial game between a generator and a critic adjudicated by a weaker LLM judge, reduces reward hacking compared to a reinforcement learning from AI feedback (RLAIF) baseline.", "We train a Gemini~2.5 Flash-class policy with a frozen, weaker Gemini~2.5 Flash Lite judge, comparing a single-player RLAIF baseline against debate."], "abstract": "We demonstrate that RL finetuning an LLM using debate, a two-player adversarial game between a generator and a critic adjudicated by a weaker LLM judge, reduces reward hacking compared to a reinforcement learning from AI feedback (RLAIF) baseline. Reward hacking is a central obstacle in RLAIF: as training progresses, the policy learns to exploit systematic errors in its AI judge, degrading task performance, a problem that worsens precisely when the judge is weaker than the policy, the setting most relevant to overseeing increasingly capable AI systems. We study mathematics tasks, where final-answer correctness is verifiable, allowing us to measure reward hacking dynamics. We train a Gemini~2.5 Flash-class policy with a frozen, weaker Gemini~2.5 Flash Lite judge, comparing a single-player RLAIF baseline against debate. While the baseline quickly hacks the judge, debate maintains judge performance throughout training, leading to a higher peak validation accuracy (45\\% performance gap recovered) that persists through many RL steps. Additional experiments show that: 1) further weakening the judge leads to faster hacking, but this can be compensated by adding an additional debate round; 2) debate incentives override prompted misalignment; 3) RL using an LLM judge has a smaller train/validation reward gap than RL from verifiable rewards; 4) learning to critique to convince the judge using ground truth labels is possible but slow. Taken together, our results are a positive update on the feasibility of debate, while highlighting that balancing multi-agent training is critical: without player constraints, adversarial training risks defaulting to critic judge-hacking. We show that critique word limits (effective up to 150 words) successfully balance the game and avoid judge hacking, though this introduces a trade-off by restricting critic expressive clarity.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-c41163f38203ce9c", "title": "Conformal Prediction for Molecular Properties under Label Shift", "authors": ["Hyeonsu Lee", "Juyeon Kim", "Erkhembayar Jadamba", "Seungjin Choi", "Hyunjin Shin"], "year": 2026, "published": "2026-08-18T11:50:38Z", "updated": "2026-08-18T11:50:38Z", "arxiv_id": "http://arxiv.org/abs/2608.17678v1", "url": "https://arxiv.org/abs/2608.17678v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["ying", "kling one of the most pervasive obstacles to applying AI in real-world drug development. By moving beyond accuracy alone to provide actionable confidence measures"], "key_results": ["Drug discovery and development underpins healthcare but remains costly and failure-prone."], "abstract": "Drug discovery and development underpins healthcare but remains costly and failure-prone. A critical bottleneck lies in predicting molecular properties such as solubility, potency, and toxicity, which directly determine whether a candidate can advance from preclinical to clinical trials. Artificial Intelligence (AI) has accelerated this process, yet its reliability is often undermined by distribution shift, as experimental conditions frequently diverge from training data. In addition, conventional point predictions provide only single-value estimates, offering limited guidance for high-stakes experimental design. We address these challenges with a conformal prediction framework tailored to label shift. By weighting conformal scores using marginal label probability ratios, our method produces statistically rigorous prediction intervals without retraining. This enables robust uncertainty quantification even when property distributions drift, directly tackling one of the most pervasive obstacles to applying AI in real-world drug development. By moving beyond accuracy alone to provide actionable confidence measures, our approach enhances the trustworthiness of AI-driven predictions. This further aligns predictive modeling with regulatory demands for transparency and uncertainty reporting and ultimately supports more reliable decision-making in billion-dollar development pipelines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-650e7d01305bf7ec", "title": "From Corpora to Co-Evolving Capabilities: Capability-Centric Data Design for Generalist Image Generation", "authors": ["Xingjian Wang", "Zhao Wang", "Taihang Hu", "Jun Zheng", "Qing Jin", "Qinye Zhou", "Zhengtao Wu", "Yongchao Du", "Zuan Gao", "Chao Lin", "Yefeng Shen", "Xiaoli Xu", "Zhengze Xu", "Hao Yan", "Yuhang Yu", "Mingzhou Zhang", "Mengting Chen"], "year": 2026, "published": "2026-08-18T17:59:01Z", "updated": "2026-08-18T17:59:01Z", "arxiv_id": "http://arxiv.org/abs/2608.18076v1", "url": "https://arxiv.org/abs/2608.18076v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Video Generation", "VLM"], "benchmarks": [], "models": [], "key_results": ["Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation."], "abstract": "Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation. A central challenge is not only how to curate each task-specific corpus, but also how to organize heterogeneous supervision according to the dependencies among generative capabilities. We present a \\textbf{capability-driven data infrastructure} that couples capability-specific supervision construction with capability-aligned curriculum scheduling. Its three specialized yet interoperable data engines build complementary relational supervision for text-image grounding, inter-image transformation, and image-knowledge association, while caption experts align T2I and editing supervision across tasks and granularities. A multi-stage curriculum jointly evolves task composition, visual-concept distribution, data quality, and image resolution along the dependency order of capability acquisition, with capability-aware evaluation closing the loop through targeted retrieval, expert construction, and gap-aware resampling. At scale, the framework curates a 440M-image T2I corpus, 120M editing pairs, and over 27M image-entity pairs. With this infrastructure, we train multimodal diffusion models at two scales from scratch, with 3B and 6B sizes respectively. We conduct quantitative evaluation on CPI-Bench, along with qualitative evaluations across diverse text-to-image and editing scenarios. Experimental results present broad visual coverage, versatile rendering, and effective transfer across generative capabilities.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-26b8c528a8e59cc5", "title": "Plug-and-Play Traffic Element Awareness for End-to-End Autonomous Driving", "authors": ["Zongzheng Zhang", "Jijun Wang", "Saining Zhang", "Shuo Wang", "Yiru Wang", "Hai Yang", "Yang Chen", "Yuwen Heng", "Hao Sun", "Anqing Jiang", "Hao Zhao"], "year": 2026, "published": "2026-08-18T17:29:51Z", "updated": "2026-08-18T17:29:51Z", "arxiv_id": "http://arxiv.org/abs/2608.18035v1", "url": "https://arxiv.org/abs/2608.18035v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation", "VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Across all paradigms and datasets, this simple integration consistently improves driving performance, demonstrating that traffic element awareness provides a robust and generalizable signal for end-to-end driving systems.", "Notably, on the challenging NAVSIM-v2 benchmark, our approach significantly improves state-of-the-art architectures and data pipelines, establishing a new state of the art."], "abstract": "Traffic elements such as traffic lights and road signs play a fundamental role in human driving decisions and should naturally influence end-to-end driving performance. However, existing end-to-end driving research predominantly focuses on dynamic road participants (e.g., vehicles and pedestrians), while the role of traffic elements remains largely unexplored. The community still lacks a systematic study quantifying their impact, largely because public datasets rarely provide structured traffic-element annotations and modern driving systems vary widely in architecture and training paradigm. In this work, we present the first systematic investigation of traffic element awareness for end-to-end autonomous driving. We construct a unified research infrastructure by augmenting multiple public driving datasets with comprehensive traffic-element annotations. To support diverse model families, we adopt a minimal and universal integration design that incorporates traffic-element signals into existing pipelines in a plug-and-play manner with negligible architectural modification. We evaluate this design across modern paradigms, including perception-prediction-planning pipelines, vision-language-action models (VLA), regression-based planners, diffusion-based policies, and trajectory-scoring frameworks, on nuScenes, NAVSIM-v1, NAVSIM-v2, and Bench2Drive. Across all paradigms and datasets, this simple integration consistently improves driving performance, demonstrating that traffic element awareness provides a robust and generalizable signal for end-to-end driving systems. Notably, on the challenging NAVSIM-v2 benchmark, our approach significantly improves state-of-the-art architectures and data pipelines, establishing a new state of the art.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-aca777dee2fa67b2", "title": "Memory Tree Guided Key Frame Querying for Efficient 3D Question Answering", "authors": ["Hsiang-Wei Huang", "Fu-Chen Chen", "Li-Wu Tsao", "Cheng-Han Lee", "Che-Chun Su", "Lu Xia", "Ronghui Peng", "Jenq-Neng Hwang", "Min Sun", "Cheng-Hao Kuo"], "year": 2026, "published": "2026-08-18T16:56:35Z", "updated": "2026-08-18T16:56:35Z", "arxiv_id": "http://arxiv.org/abs/2608.18009v1", "url": "https://arxiv.org/abs/2608.18009v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Memory / RAG", "VLM"], "benchmarks": [], "models": ["GPT-4o", "LLaVA-OneVision-7B", "ying"], "key_results": ["On OpenEQA, our method improves the LLM-Match of GPT-4o by 17.4%, LLaVA-OneVision-7B by 5.8%, outperforms existing visual search methods."], "abstract": "Answering questions accurately and efficiently in embodied scenarios presents significant challenges due to limited computational and memory resources for Vision Language Model (VLM) inference. Existing methods adopt visual search key frame retrieval method to select critical question-related key frames for VLM input. However, visual search methods are inefficient because they require visual search among thousands of video frames for each individual user query. In this work, we propose a memory tree guided key frame selection paradigm for efficient 3D question answering in embodied scenarios. Our method leverages a compact and reusable 3D scene representation, termed MemTree3D, which supports real-time online construction leveraging camera 6-DoF poses. MemTree3D captures multi-level 3D scene information, enabling a Large Language Model to efficiently query and retrieve question-relevant key frames through our scoring-based frame selection without reprocessing the entire video stream. On OpenEQA, our method improves the LLM-Match of GPT-4o by 17.4%, LLaVA-OneVision-7B by 5.8%, outperforms existing visual search methods. Our code is available at https://github.com/hsiangwei0903/MemTree3D", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-d7fe65fbd986be34", "title": "LinCa: Accelerating Diffusion Models via Learnable Decomposed Feature Caching", "authors": ["Jinshan Liu", "Haoran Qin", "Xiaobing Tu", "Jiacheng Liu", "Jiahui Hu", "Zhengan Yan", "Yukun Xie", "Kerui Shen", "Jinkui Ren", "Yuqi Lin", "Xiantao Zhang", "Linfeng Zhang"], "year": 2026, "published": "2026-08-18T16:26:21Z", "updated": "2026-08-18T16:26:21Z", "arxiv_id": "http://arxiv.org/abs/2608.17973v1", "url": "https://arxiv.org/abs/2608.17973v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": ["Qwen-Image", "FLUX", "HunyuanVideo"], "key_results": ["Diffusion models have achieved remarkable success in image and video generation, yet the high computational cost of iterative sampling remains a critical bottleneck for practical deployment.", "Experiments on FLUX, Qwen-Image, and HunyuanVideo demonstrate that LinCa, with less than 0.2% additional parameters, significantly outperforms existing methods and maintains near-lossless quality at 5-7x speedup."], "abstract": "Diffusion models have achieved remarkable success in image and video generation, yet the high computational cost of iterative sampling remains a critical bottleneck for practical deployment. Feature caching has emerged as a promising acceleration paradigm by reusing or predicting intermediate features across timesteps. However, existing training-free methods apply uniform prediction strategies that cannot adapt to the heterogeneous feature dynamics, causing significant quality degradation under high acceleration ratios. We propose LinCa, a feature caching framework based on learnable invertible networks. LinCa decomposes cached features into sub-components with distinct continuity properties via a lightweight invertible network and applies differentiated prediction orders matched to each component. The strict invertibility guarantees lossless reconstruction back to the original feature space, forming a unified Decompose-Predict-Reconstruct pipeline. By training separate predictors for different models and timestep segments, LinCa adapts to heterogeneous feature dynamics. Experiments on FLUX, Qwen-Image, and HunyuanVideo demonstrate that LinCa, with less than 0.2% additional parameters, significantly outperforms existing methods and maintains near-lossless quality at 5-7x speedup. Code: https://github.com/QHR69/LinCa", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-2da5af8b3accb805", "title": "An Omitted Mode Is a Rare Rule: The Sampling-Verification Danger Law in Continuous Code World Models", "authors": ["Javier Aguilar Martín"], "year": 2026, "published": "2026-08-18T16:09:51Z", "updated": "2026-08-18T16:09:51Z", "arxiv_id": "http://arxiv.org/abs/2608.17956v1", "url": "https://arxiv.org/abs/2608.17956v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "eess.SY"], "pillar": "Multimodal / World Models", "methods": ["World Model"], "benchmarks": [], "models": ["GPT-5.x"], "key_results": ["In the Code World Model paradigm an LLM synthesizes an executable world model that a classical planner searches, and the model is accepted when it reproduces sampled transitions."], "abstract": "In the Code World Model paradigm an LLM synthesizes an executable world model that a classical planner searches, and the model is accepted when it reproduces sampled transitions. We ask what that acceptance certifies in continuous control. We define the pipeline's danger as an expected risk and isolate its exact factor: the probability that N i.i.d. gate rollouts all miss a critical event of probability r is exactly (1-r)^N; an independent acceptance sample adds its budget to the exponent. On three hybrid instruments the accepted mode-blind model is exploited: the planner is pinned at the mode boundary at a regret of nearly the whole attainable return. We prove a localization budget, valid at boundary points: models with Lipschitz constant at most L differing by eta at a point disagree above tolerance eps on a region of volume at least kappa((eta-eps)/L)^(d+m); the discontinuous reset modes studied pay no such budget. With real LLM synthesis, GPT-5.x repairs an omitted 1D clamp in 105 of 111 mode-containing draws -- every attempt exact on 50 of 56 instrument-stream blocks (95% CI [0.781, 0.960]). On 2D regions no artifact recovers the rule (0/156); eight targeted interventions leave the failure in place, and positive controls locate it: a located rule is not induced, while given form and location the constants follow exactly. A version-space certificate proves identification is class-relative: at the widest dose the declared fit succeeds in 20/20 blocks and every sample-consistent circle is within tolerance in 18/20. We prove a class of entry rules exactly consistent with every sample yet harmless at play, so identifiability is a measurable property of the instrument. Re-scoring all 1034 artifacts on independent samples confirms acceptance certifies sample consistency and no more: where the gate is provably informative it covers about two percent of the exploited planner's queries.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-9099b319cd0241fc", "title": "PerFact: Perception-Derived Fact Prompting for 3D Brain MRI Report Generation", "authors": ["Jianyu Sun", "Zhenxuan Zhang", "Guang Yang", "Peter J. Lally"], "year": 2026, "published": "2026-08-18T15:47:45Z", "updated": "2026-08-18T15:47:45Z", "arxiv_id": "http://arxiv.org/abs/2608.17926v1", "url": "https://arxiv.org/abs/2608.17926v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["In a controlled study that fixes the backbone, data split, target reports, and adaptation while varying only the injected grounding, perception-derived facts outperform retrieved prior reports, retrieval becomes redundant once facts are present, and end-to-end predicted facts remain effective without any ground-truth annotation at inference."], "abstract": "Radiology report generation has matured almost entirely on 2D chest radiographs, where the default route to better reports is a larger backbone or a pre-training one on medical data. We revisit that assumption on 3D multi-sequence brain MRI, a volumetric multi-disease regime, and find that the model is not the lever. Zero-shot medical and radiology vision-language models transfer poorly to brain MRI, with chest radiograph specialists failing most conspicuously, and five backbones fine-tuned identically across three model families and an order of magnitude in scale differ only marginally. What determines the quality of the report is the information injected into the prompt. We delegate perception to upstream 3D segmentation and classification, serialize their outputs into a structured fact sentence, and prompt a LoRA-adapted vision-language model with it; we call this \\textbf{PerFact}. In a controlled study that fixes the backbone, data split, target reports, and adaptation while varying only the injected grounding, perception-derived facts outperform retrieved prior reports, retrieval becomes redundant once facts are present, and end-to-end predicted facts remain effective without any ground-truth annotation at inference. The residual gap between predicted and oracle facts is explained by the granularity of the facts rather than by the generator. Closed-ended visual question answering comes at no measurable cost to report quality, though the grounding source has little effect on it. On 3D brain MRI, grounding information, not model choice, is the dominant controllable factor in report quality.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-79bb9f49e2f2762e", "title": "BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models", "authors": ["Liubov Chubarova", "Alexandra Kuleshova", "Daniil Volkov", "Kirill Sultanov", "Alexey Zaytsev"], "year": 2026, "published": "2026-08-18T15:29:09Z", "updated": "2026-08-18T15:29:09Z", "arxiv_id": "http://arxiv.org/abs/2608.17895v1", "url": "https://arxiv.org/abs/2608.17895v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["Gemini 3.1 Pro and Qwen3.5-397B", "Qwen3.5-397B"], "key_results": ["While Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, professional documents remains incompletely evaluated."], "abstract": "While Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, professional documents remains incompletely evaluated. Existing benchmarks emphasize information extraction, require external domain knowledge, or cover professional documents only as one of many settings. They are also largely English- or Chinese-centric, leaving other languages and Russian, in particular, substantially underrepresented. To address these limitations, we introduce BEAR-Bench (Bilingual Enterprise and Academic Reasoning), a self-contained, complex English-and-Russian benchmark comprising 1000 human-annotated questions based on text-rich business and scientific documents. We evaluate 16 proprietary and open-weight MLLMs, including Gemini 3.1 Pro and Qwen3.5-397B, on BEAR-Bench and observe clear headroom even for the strongest systems. Finally, we use the resulting model outputs to compare existing hallucination detection methods, evaluating not only how often models fail on BEAR-Bench but also how reliably those failures can be identified.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-577b2b88892659a6", "title": "Vision-Language Models for Analog Gauge Reading: An Empirical Study of Specialization, Transfer and Reliability", "authors": ["Abdul Mueez", "Aaditya Baranwal", "Junior Chaj-Mejia", "Guneet Bhatia", "Jason T. Voelker", "Shruti Vyas"], "year": 2026, "published": "2026-08-18T12:47:14Z", "updated": "2026-08-18T12:47:14Z", "arxiv_id": "http://arxiv.org/abs/2608.17723v1", "url": "https://arxiv.org/abs/2608.17723v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Quantization", "Safety / Jailbreak", "VLM"], "benchmarks": [], "models": ["Qwen2.5-VL-7B-Instruct"], "key_results": ["Analog gauges remain common in industrial environments where manual inspection is costly or hazardous."], "abstract": "Analog gauges remain common in industrial environments where manual inspection is costly or hazardous. The engineering application addressed here is direct numerical reading of single-target analog-gauge images, while the artificial-intelligence contribution is a systematic evaluation of specialization, transfer, robustness and reliability for a general-purpose vision-language model (VLM) without an explicit pointer-segmentation and geometric-reading pipeline. The Qwen2.5-VL-7B-Instruct model is evaluated using zero-shot prompting, in-context learning (ICL) and parameter-efficient fine-tuning with Quantized Low-Rank Adaptation (QLoRA) on a public synthetic dataset, a video-derived Pressure Gauge dataset and a proprietary industrial dataset. All fine-tuning experiments use a fixed 20-epoch protocol with the final epoch used for analysis; separate models with and without supplied gauge ranges remove prompt-setting confounds. The primary metric is range-normalized mean percentage error (MPE). The best fine-tuned MPE values are 2.39% on the synthetic dataset, with a 95% bootstrap confidence interval (CI) of 1.43-3.90%; 2.61% on the Pressure Gauge dataset, with a CI of 1.66-3.80%; and 4.43% on the proprietary industrial dataset, with a CI of 2.31-7.14%. Leave-one-dataset-out experiments reveal substantial transfer degradation on held-out synthetic and proprietary data, while robustness tests identify Gaussian blur as the strongest tested corruption. Reliability analysis shows that high-confidence errors remain possible, motivating abstention and independent validation in safety-critical use. These results support QLoRA-specialized VLMs for direct single-gauge reading but not yet a deployment-ready plant-monitoring pipeline.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-916c5463455522ae", "title": "MemCatalyst: Amplifying Data Auditing on Vision-Language Models via Data Poisoning", "authors": ["Xukun Luan", "Jinyan Liu", "Yuhui Gong", "Yuanguo Bi", "Bing Hu", "Xuesong Li", "Di Wang"], "year": 2026, "published": "2026-08-18T12:46:52Z", "updated": "2026-08-18T12:46:52Z", "arxiv_id": "http://arxiv.org/abs/2608.17722v1", "url": "https://arxiv.org/abs/2608.17722v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Vision-Language models (VLMs) achieve outstanding performance largely due to the amount of training data available on the internet.", "Extensive evaluations using five state-of-the-art data audits on two prominent VLMs demonstrate that MemCatalyst markedly enhances MI AUC scores with a minimal budget of poisoned samples, while maintaining a negligible impact on model performance."], "abstract": "Vision-Language models (VLMs) achieve outstanding performance largely due to the amount of training data available on the internet. At the same time, data holders (e.g., artists) urgently need to determine whether their data has been used for model training without authorization, which concerns both intellectual property rights and personal privacy. Data auditing, particularly through membership inference (MI), has attracted attention as a direct tool. This work proposes MemCatalyst, a set of data poisoning tools, aiming to amplify the data auditing performance on VLMs. MemCatalyst employs two strategies: Poisoning Text (PT) and Poisoning Image (PI). MemCatalyst forces VLMs to over-learn specific inconsistencies between image features and textual semantics during training, thereby increasing their susceptibility to membership information auditing. Crucially, the transferability of poisoned samples across different VLM architectures is demonstrated to be effective in the black-box setting. Extensive evaluations using five state-of-the-art data audits on two prominent VLMs demonstrate that MemCatalyst markedly enhances MI AUC scores with a minimal budget of poisoned samples, while maintaining a negligible impact on model performance.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-d2f774bb267ef447", "title": "Magnitude-Direction Decoupling for Fast Video Generation with Flow Matching Models", "authors": ["Haonan Xu", "Feiyang Chen", "Songkui Chen", "Hongpeng Pan", "Zhefeng Wang", "Xinyu Duan", "Baoxing Huai", "Yang Yang"], "year": 2026, "published": "2026-08-18T12:12:21Z", "updated": "2026-08-18T12:12:21Z", "arxiv_id": "http://arxiv.org/abs/2608.17695v1", "url": "https://arxiv.org/abs/2608.17695v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Flow matching models for video generation achieve impressive performance but suffer from high computational overhead due to iterative denoising.", "Moreover, MDD further reduces inference costs by reusing magnitude information under classifier-free guidance (CFG).", "Experiments show that MDD outperforms existing acceleration methods, delivering promising speedups (e.g., up to 2.95x on Wan2.1) while preserving high visual fidelity and content richness."], "abstract": "Flow matching models for video generation achieve impressive performance but suffer from high computational overhead due to iterative denoising. In fact, the original model is not necessary for all denoising steps, allowing some steps to use lightweight alternatives for faster sampling. However, directly using caching or lightweight models can deviate from the original denoising trajectory, resulting in suboptimal performance. Through empirical analysis, we find that lightweight models can robustly capture the magnitude components of the original model's output, while caching provides reliable directional guidance. Building on this insight, we propose the Magnitude-Direction Decoupling (MDD) method, which adaptively employs a direction-calibrated lightweight model as a substitute for the original model to accelerate inference and effectively correct deviations in the denoising trajectory. Moreover, MDD further reduces inference costs by reusing magnitude information under classifier-free guidance (CFG). As a result, MDD offers a more reliable and lightweight solution to accelerate sampling. Experiments show that MDD outperforms existing acceleration methods, delivering promising speedups (e.g., up to 2.95x on Wan2.1) while preserving high visual fidelity and content richness.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-014a7f75fdefb990", "title": "Multi-turn Conversational AI from Text to Multimodal Interaction: Data, Models, Evaluation, and Open Challenges", "authors": ["Syeda Faiza Ahmed", "Zien Sheikh Ali", "Hunzalah Hassan Bhatti", "Firoj Alam", "Shammur Absar Chowdhury"], "year": 2026, "published": "2026-08-18T10:14:21Z", "updated": "2026-08-18T10:14:21Z", "arxiv_id": "http://arxiv.org/abs/2608.17605v1", "url": "https://arxiv.org/abs/2608.17605v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.SD"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Memory / RAG", "Deep Research", "VLM"], "benchmarks": [], "models": [], "key_results": ["Conversational AI is moving beyond isolated text prompts toward sustained, multimodal interaction."], "abstract": "Conversational AI is moving beyond isolated text prompts toward sustained, multimodal interaction. In real conversations, users clarify goals, revise requests, interrupt responses, switch topics, and introduce new evidence while expecting systems to preserve context across turns. This makes multi-turn dialogue a distinct challenge requiring systems to maintain and update memory, ground responses across modalities, tools, and external knowledge, and adapt across languages and cultures. This study reviews multi-turn conversational AI across text-only dialogue, AudioLLMs and speech-native systems, multimodal and omni-modal systems, and tool-augmented agents. We organize the literature around datasets and benchmarks, modeling paradigms, training strategies, evaluation setups, and cross-cutting challenges. Our analysis shows that support for multiple modalities has advanced faster than the ability to sustain coherent interaction across a session. Despite stronger capabilities to perceive, speak, and act across modalities, current systems still struggle with persistent memory, cross-turn grounding, full-duplex interaction, robust evaluation, and cultural alignment. We conclude with a research agenda for systems that can remember, revise, ground, speak, listen, act, and adapt across turns, modalities, and cultures. (https://github.com/faiza-sfa/multiturn-conversational-ai-survey)", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-6182a3b8fc707c90", "title": "Auditing Exposure to Harmful Content on TikTok using Multimodal Language Models: A Cross-National, Age-Stratified Study", "authors": ["Hamidreza Saffari", "Francesco Pierri"], "year": 2026, "published": "2026-08-18T09:49:02Z", "updated": "2026-08-18T09:49:02Z", "arxiv_id": "http://arxiv.org/abs/2608.17583v1", "url": "https://arxiv.org/abs/2608.17583v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Multimodal / World Models", "methods": ["Safety / Jailbreak", "VLM"], "benchmarks": [], "models": ["Gemini 2.5 Flash with eight sampled frames plus text performs best"], "key_results": ["To scale annotation, we validate four multimodal LLMs against native-speaker labels on a 300-video reference set."], "abstract": "Online video platforms can expose young users to harmful content, but independent audits remain difficult because video annotation is costly and moderation judgments vary across languages. We audit TikTok in France, Italy, and Sweden with sockpuppet accounts representing four age personas (13, 16, 19, 40), collecting 36,971 videos from passive For-You-page scrolling and active sessions that scroll, search for harm keywords, and scroll again. To scale annotation, we validate four multimodal LLMs against native-speaker labels on a 300-video reference set. Gemini 2.5 Flash with eight sampled frames plus text performs best (aggregate kappa = 0.42), at half the per-call cost of native-video upload, and we apply it to a 10% sample for approximately \\$50 in total API spend across both modalities. Keyword search returns 35-56% harmful content, a 1.5-7.5x increase over the scrolling baseline in ten of twelve country-age combinations; the spike is temporary and flattens the age differences observed in France and Sweden. Under passive scrolling, Italy has the highest harm rate at every age, with Italian age-19 reaching 48.6%. Overall, MLLM-based auditing offers a scalable approach for cross-national youth-safety audits, while provider safety filters (1.1% refusal rate) under-count the most explicit harms.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-37fcc1e1e55f355d", "title": "Where a New Concept Must Enter: Entry Point Gates Cross-Task Usability in Unified Multimodal Models", "authors": ["Zongyang Qiu", "Yihan Wu", "Kaixuan Fan", "Bo Li", "Hui Xiong"], "year": 2026, "published": "2026-08-18T09:23:35Z", "updated": "2026-08-18T09:23:35Z", "arxiv_id": "http://arxiv.org/abs/2608.17564v1", "url": "https://arxiv.org/abs/2608.17564v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "Computer Use", "VLM"], "benchmarks": [], "models": [], "key_results": ["Exploiting the rule, a mid-stack alignment objective acquires the concept for a $0.1\\%$ relative loss of the model's general text-to-image ability, against $41\\%$ for the standard generative route."], "abstract": "Unified multimodal models (UMMs) are motivated by the hope that understanding and generation reinforce each other but controlled ablations repeatedly find that adding a generation objective leaves understanding flat. Joint-training studies cannot settle the disagreement: with overlapping supervision, a gain cannot be attributed to the architecture rather than the data. To further investigate the relationship between the two directions in UMMs, we separate them by construction. A novel visual entity, a rendered 3D asset paired with a pseudo-word screened for absence from the frozen model's behavior, is bound through exactly one task direction, and the untrained direction is then measured. We find that the channel is real in both directions, but the directions differ in kind: generation training installs a name the model can only match among candidates; understanding training installs one it can also produce. What governs cross-task usability is where the binding enters the shared computation. An alignment probe predicts export across 36 configurations (Spearman $ρ= +0.68$). That objective's alignment term, maximized in closed form over activations with every weight frozen, makes a concept drawable when injected at layer 7 of 28 and is indistinguishable from the base model from layer 14 on, while the weight-based version of the same edit peaks at layers 10-14. In an observational series of four models, this window appears only where the understanding pathway is a semantic vision encoder, suggesting that unified weights are not enough: the two directions must share a semantic format at the entry point. Exploiting the rule, a mid-stack alignment objective acquires the concept for a $0.1\\%$ relative loss of the model's general text-to-image ability, against $41\\%$ for the standard generative route. Our code is at https://github.com/Zane-ZYQiu/entry-point-umm.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-0704506ffca193e2", "title": "Toward the Optimal Regret-Instability Trade-off in Multi-Armed Bandits", "authors": ["Kaifei Wang", "Yinyu Ye", "Han Zhong"], "year": 2026, "published": "2026-08-18T14:38:47Z", "updated": "2026-08-18T14:38:47Z", "arxiv_id": "http://arxiv.org/abs/2608.17841v1", "url": "https://arxiv.org/abs/2608.17841v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG", "math.OC", "math.ST"], "pillar": "Quant × AI", "methods": ["Deep Research"], "benchmarks": ["MATH"], "models": [], "key_results": ["Multi-armed bandit algorithms are evaluated by regret, yet comparable regret can coexist with different allocations across independent runs."], "abstract": "Multi-armed bandit algorithms are evaluated by regret, yet comparable regret can coexist with different allocations across independent runs. We study the trade-off between worst-case regret $\\mathcal{R}_{K,T}$ and instability $\\mathcal S_{K,T}$, defined as the largest standard deviation of a terminal pull count, for $K$ arms and $T$ rounds. We prove the finite-time lower bound $\\mathcal R_{K,T}\\mathcal S_{K,T}\\ge C T^{3/2}$, where $C$ is independent of $K$ and $T$, under a finite-time regret condition and without the regularity assumptions imposed in the prior asymptotic analysis. We also introduce Stabilized Lower-Envelope UCB (\\textup{\\textsc{SLE-UCB}}), a new tunable algorithm combining a running lower-envelope index with a decreasing pull-count stabilizer. \\textup{\\textsc{SLE-UCB}} satisfies $\\mathcal R_{K,T}\\mathcal S_{K,T}=O(T^{3/2}\\log K)$, with an implicit constant independent of $K$ and $T$, matching the lower bound exactly in $T$ and within a logarithmic factor in $K$. To prove the instability bound, we develop a new offline top-prefix representation that removes path dependence from online decisions. Together with single-reward perturbations and the Efron--Stein inequality, this representation controls pull-count variance. Thus, regret and instability depend reciprocally on $K$, while their product has no polynomial dependence on $K$. These results resolve the open question raised in the literature concerning the sharp arm-dependent regret--instability frontier.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-0cb2dc467029c620", "title": "Spatially explicit feature importance for building height estimation using research-access high-resolution SAR and optical sensors", "authors": ["Guilherme Iablonovski", "Pierre-Louis Frison", "Tatiana Silva da Silva"], "year": 2026, "published": "2026-08-18T14:20:53Z", "updated": "2026-08-18T14:20:53Z", "arxiv_id": "http://arxiv.org/abs/2608.17822v1", "url": "https://arxiv.org/abs/2608.17822v1", "primary_category": "physics.soc-ph", "categories": ["physics.soc-ph", "cs.LG", "stat.ML"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["phically"], "key_results": ["To account for the spatial autocorrelation in the training set, features from all sources are integrated in a geographically weighted random forest model, returning an RMSE of 5.34 m and R2 of 0.756 against a LiDAR reference dataset."], "abstract": "Accurate building height information at the individual footprint scale is essential for material stock accounting and post-disaster damage assessments yet remains difficult to obtain at city scale in the Global South where airborne LiDAR coverage is rare and commercial very high-resolution imagery is cost-prohibitive or unavailable. While recent works have demonstrated building height estimation using freely available Sentinel imagery, the resolution ceiling of resulting products is still coarse for material stock analysis. This study incorporates products derived from data freely accessible under scientific research licenses, TerraSAR-X StripMap and PlanetScope, alongside Sentinel-1 to predict building heights in a large city in Brazil. To account for the spatial autocorrelation in the training set, features from all sources are integrated in a geographically weighted random forest model, returning an RMSE of 5.34 m and R2 of 0.756 against a LiDAR reference dataset. Local feature importance showed predictor dominance to vary consistently across intra-urban contexts, with footprint geometry dominating for low-rise buildings, shadow-derived height for taller and more isolated structures, and spectral reflectance for the tallest buildings in the set. Sentinel-1 backscatter and InSAR occupy complementary spatial niches, with no single sensor uniformly preferable across the set. Results provide optioneering guidance and insight over satellite-derived products predictive relevance in distinct contexts, which global machine learning or neural network models cannot offer.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-66767039dc4eb97b", "title": "Self-Consistent Adjoint Policy Iteration for Constrained Dynamic Portfolio Choice", "authors": ["Jeonggyu Huh", "Yeoneung Kim", "Seungwon Jeong"], "year": 2026, "published": "2026-08-18T14:05:51Z", "updated": "2026-08-18T14:05:51Z", "arxiv_id": "http://arxiv.org/abs/2608.17808v1", "url": "https://arxiv.org/abs/2608.17808v1", "primary_category": "math.OC", "categories": ["math.OC", "q-fin.CP", "q-fin.PM"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": ["MATH"], "models": ["yields"], "key_results": ["Shifted-adjoint cancellation controls the adjoint--HJB Hamiltonian-gradient discrepancy by the policy-improvement residual.", "A theorem-matched audit yields a maximal 95% upper endpoint of 0.074 against the required 0.75 threshold.", "In a three-factor, fifty-asset design, current-policy re-evaluation outperforms matched pooled refinement under both evaluation laws."], "abstract": "We develop simulation-based policy iteration for continuous-time portfolio choice with predictable returns and convex constraints. Each outer step re-evaluates a fixed-latent OL-BPTT adjoint after deployment and solves the constrained update. Shifted-adjoint cancellation controls the adjoint--HJB Hamiltonian-gradient discrepancy by the policy-improvement residual. For CRRA portfolios, exact HJB policy iteration identifies the optimal reduced value factor, while population OL-BPTT iteration converges globally under an occupation-measure relative-error condition. A theorem-matched audit yields a maximal 95% upper endpoint of 0.074 against the required 0.75 threshold. In a three-factor, fifty-asset design, current-policy re-evaluation outperforms matched pooled refinement under both evaluation laws.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-547b59f464679d81", "title": "When to Review: Spaced Repetition for Continual Pre-Training of Language Models", "authors": ["Alankar Atreya", "Devesh Batra", "Yoages Kumar Mantri", "Geremy Bantug", "Greig A Cowan", "Raad Khraishi"], "year": 2026, "published": "2026-08-18T08:51:47Z", "updated": "2026-08-18T08:51:47Z", "arxiv_id": "http://arxiv.org/abs/2608.17530v1", "url": "https://arxiv.org/abs/2608.17530v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["On temporally separated Wikipedia and code corpora, SRT improves the stability-plasticity trade-off, recovering 5 to 37 percentage points of old-knowledge accuracy lost by naive continual pre-training across model scales while preserving or improving new-knowledge acquisition."], "abstract": "Continual pre-training of large language models must acquire new information without erasing old knowledge. Existing replay methods often choose a global old/new mixture and sample uniformly, ignoring that examples differ in how quickly they are forgotten. We formulate continual pre-training as adaptive review scheduling: the training loop should decide not only how much history to replay, but which examples should return at each step. We introduce Spaced Repetition Training (SRT), a continual learning framework inspired by cognitive science, which schedules sample-rehearsal using the SuperMemo-2 (SM-2) algorithm. SRT maintains per-example review state, maps per-example perplexity to a recall-quality signal, and schedules historical examples for retention and new examples for consolidation while leaving the model, objective, and optimizer unchanged. On temporally separated Wikipedia and code corpora, SRT improves the stability-plasticity trade-off, recovering 5 to 37 percentage points of old-knowledge accuracy lost by naive continual pre-training across model scales while preserving or improving new-knowledge acquisition. At larger scale, SRT preserves broad benchmark performance that naive continual pre-training and uniform replay substantially degrade. Experiments with vision and tabular data further suggest that the scheduling principle extends beyond language when paired with an appropriate recall signal.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-7603d09f173de83e", "title": "Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Requirements", "authors": ["Zhi Zheng", "Rongsheng Chen", "Yunpeng Ba", "Zhenkun Wang", "Yee Whye Teh", "Wee Sun Lee"], "year": 2026, "published": "2026-08-18T03:03:53Z", "updated": "2026-08-18T03:03:53Z", "arxiv_id": "http://arxiv.org/abs/2608.17310v1", "url": "https://arxiv.org/abs/2608.17310v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO", "Test-time Scaling", "Multi-Agent", "Memory / RAG"], "benchmarks": ["WebArena"], "models": ["Qwen-3.5-27B", "yielding"], "key_results": ["To improve the exploration--adaptation trade-off, Agentic ESOpt further introduces a cosine decay schedule of the perturbation scale $σ$.", "On WebArena-Lite, full-parameter optimization of Qwen-3.5-27B improves the No Skill baseline by 6.69%."], "abstract": "Reinforcement Learning (RL) has been promising in single-turn LLM fine-tuning. However, long-horizon agentic reasoning introduces increasingly branching interactions and sparse rewards, exposing several limitations of RL: its heavyweight backpropagation-based training stack makes it impractical to fine-tune larger LLMs, and longer-horizon trajectories make credit assignment in RL substantially harder. This paper argues that evolution strategies (ES) can be a better choice for fine-tuning long-horizon LLM agents. Compared with agentic RL, ES offers three key advantages: 1) Model Scalability: ES enables full-parameter optimization with only minimal, inference-level GPU memory, making it possible to fine-tune large LLMs. 2) Flexibility: its lightweight, black-box feedback interface makes ES fine-tuning easy to compose with prompt-space evolution (e.g., skill optimization & test-time compute); and 3) Long-Horizon Scalability: ES performs trajectory-level parameter attribution without decomposing rewards across horizons, yielding better scalability than Agentic RL as the horizon length grows. Based on this insight, we propose Agentic ESOpt, a full-parameter agentic fine-tuning framework tailored to flexible parameter--context co-evolution. At each step, Agentic ESOpt samples perturbations around the current LLM parameters, evaluates the resulting agents with rewards, and applies an online reward-weighted update. To improve the exploration--adaptation trade-off, Agentic ESOpt further introduces a cosine decay schedule of the perturbation scale $σ$. On WebArena-Lite, full-parameter optimization of Qwen-3.5-27B improves the No Skill baseline by 6.69%. In test-time automatic heuristic design, Agentic ESOpt performs online prompt--parameter co-evolution, improving its matched baseline in 28 of 36 settings.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-436b12cc7f912e69", "title": "Temporal Leakage in Financial News NLP: A Multi-Architecture Audit with a Regime-Specific M&A Signal", "authors": ["Chenhao Xue", "Raslen Guesmi", "Siwei Feng", "Yucheng Gong", "Jacob Xavier Sundram", "Jordan Pang", "Lan Wang", "Julian Kaljuvee"], "year": 2026, "published": "2026-08-18T00:23:25Z", "updated": "2026-08-18T00:23:25Z", "arxiv_id": "http://arxiv.org/abs/2608.17223v1", "url": "https://arxiv.org/abs/2608.17223v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["Llama-3", "Qwen2.5", "v3-large", "BERT"], "key_results": ["Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train-test split is chronological or random, i.e., on temporal leakage."], "abstract": "Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train-test split is chronological or random, i.e., on temporal leakage. We audit this dependence on a 49,799-article corpus across 16 feature-model combinations spanning TF-IDF, MiniLM, FinBERT, and fine-tuned RoBERTa-large / DeBERTa-v3-large, plus separate zero/few-shot and LoRA probes of Llama-3 and Qwen2.5 LLMs: random splits inflate MCC by $1.1\\times$ to $6.5\\times$, tracking model capacity and feature richness, and end-to-end FinBERT fine-tuning re-amplifies rather than closes the gap (size-matched ratio $1.75\\times$). Conditioning on event type, mergers and acquisitions (M&A) is the only audited category with a positive locked-test signal under near-temporal chronological evaluation (TF-IDF MCC $= 0.138$ train-only, $0.068$ under train$\\cup$val refit; 10,000-permutation $p < 10^{-3}$); the signal does not transfer to FNSPID's 2009-2020 U.S. corpus, localising the headline to our 2024-2025 European-tilted M&A semantics rather than a universal predictor. Three independent role labellers converge on acquirer-tagged articles as the signal locus, a power-limited qualitative convergence rather than a hypothesis-tested asymmetry. Chronological splitting plays for financial NLP the role characteristics-purging plays for asset pricing: it strips the predictable, stale component of news and leaves a residual that is small, event-localized, and lexically shallow. We advocate leakage audits as a required disclosure for financial-NLP benchmarks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-244f65afb517acbf", "title": "Picture the Epsilon: Pursuing Identity-Level Privacy Guarantees for Images", "authors": ["Arman Zareian Jahromi", "Vishnu Bondalakunta", "Mohammad Akbar Bin Shah", "Naimul Haque", "Shuangqing Wei", "George T. Amariucai"], "year": 2026, "published": "2026-08-17T21:30:06Z", "updated": "2026-08-17T21:30:06Z", "arxiv_id": "http://arxiv.org/abs/2608.17147v1", "url": "https://arxiv.org/abs/2608.17147v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["Image-to-image face generators are widely used, and visual dissimilarity between their outputs and source images is sometimes treated as evidence of privacy."], "abstract": "Image-to-image face generators are widely used, and visual dissimilarity between their outputs and source images is sometimes treated as evidence of privacy. Auditing whether these systems satisfy formal identity-level (epsilon, delta)-differential privacy requires choosing among several distinct routes for converting embedding-space observations into estimates or bounds on the differential privacy parameter epsilon. We present a comparative study of four such audits applicable to pre-trained, black-box face generators: a Gaussian-mechanism reading of per-identity sensitivity (GaussMech); a per-dimension kernel-density log-ratio aggregated by basic composition (KDE-LR); an analytical population-level lower bound on pure-DP epsilon derived from the maximum mean discrepancy via the total variation distance (MMD-TV); and a hypothesis-testing evaluation of a cross-validated classifier's out-of-fold ROC (ROC-HT). For each method we make explicit its assumptions, hyperparameter dependence, finite-sample limitations, and the regime in which its epsilon estimate is informative. Applied to FaceFusion and InstantID across multiple identity encoders and reference datasets, the audits consistently reveal substantial identity distinguishability while reporting markedly different epsilon estimates that reflect each method's distinct assumptions and finite-sample treatment. In this high-distinguishability regime, the experiments do not support a reliable ranking of the four methods. Their relative trade-offs should be evaluated on partially private mechanisms, which we identify as the natural next study. The resulting framework places these audits in a shared identity-level audit setting and clarifies how their assumptions and finite-sample treatments shape the resulting differential privacy estimates.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-a419ee663be2b2df", "title": "Dynamic Regime-Aware Conformal Calibration for Reliable Economic Forecast Intervals under Multiple Distribution Shifts", "authors": ["Bogdan Oancea"], "year": 2026, "published": "2026-08-17T19:36:15Z", "updated": "2026-08-17T19:36:15Z", "arxiv_id": "http://arxiv.org/abs/2608.17079v1", "url": "https://arxiv.org/abs/2608.17079v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["We evaluate DRACP against six baselines on 48 real forecasting series covering euro-area and EU-27 HICP inflation, US macroeconomic and energy indicators, and daily financial series.", "Recent online methods (FACI, strongly-adaptive online conformal prediction and conformal PID) were verified against the authors' implementations.", "DRACP is not the most efficient method: strongly-adaptive online conformal prediction achieves the best interval score and intervals about 20% narrower."], "abstract": "Conformal prediction provides distribution-free prediction intervals but relies on exchangeability, an assumption often violated in economic forecasting because of covariate shift, concept drift, local heterogeneity and latent regimes. We propose Dynamic Regime-Aware Conformal Prediction (DRACP), which combines density-ratio, localized kernel and probabilistic regime-aware weighting with a self-tuning online significance controller in a unified weighted conformal calibration framework. We distinguish three theoretical results: finite-sample validity under oracle importance weights, a coverage-gap bound for estimated weights with rates in effective sample size, and deterministic or regret guarantees for the online controller. We evaluate DRACP against six baselines on 48 real forecasting series covering euro-area and EU-27 HICP inflation, US macroeconomic and energy indicators, and daily financial series. Recent online methods (FACI, strongly-adaptive online conformal prediction and conformal PID) were verified against the authors' implementations. DRACP is not the most efficient method: strongly-adaptive online conformal prediction achieves the best interval score and intervals about 20% narrower. Instead, DRACP provides the most reliable calibration, achieving coverage closest to the nominal 0.90 (0.890), never falling below 0.80 on any series, maintaining the best coverage at all forecast horizons, and performing best during the 2021-2023 inflation surge. The strongly-adaptive method undercovers on 20 of 48 series versus 10 for DRACP. DRACP therefore offers a principled trade-off between calibration and efficiency, favoring reliable coverage when prediction intervals must satisfy coverage standards. An ablation study shows that the online controller and conditional-scale normalization provide most of the performance gain, whereas the weighting components make a smaller contribution.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-19T03:25:16.672996+00:00"}
{"id": "evg-d2870996e7a74995", "title": "VIALS: A Benchmark for Visual Interpretation of Artifacts in the Life Sciences", "authors": ["Elaine Lau", "Thanuka Udumulla", "Lee Izhaki-Tavor", "Francisco Guzmán", "Nicholas Magazine", "Jonas Mueller"], "year": 2026, "published": "2026-08-21T17:59:26Z", "updated": "2026-08-21T17:59:26Z", "arxiv_id": "http://arxiv.org/abs/2608.21357v1", "url": "https://arxiv.org/abs/2608.21357v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["In professional life sciences workflows, scientists routinely interpret visual artifacts (gel blots, microscopy images, plasmid maps, flow cytometry plots, molecular structures, ...) to inform research decisions."], "abstract": "In professional life sciences workflows, scientists routinely interpret visual artifacts (gel blots, microscopy images, plasmid maps, flow cytometry plots, molecular structures, ...) to inform research decisions. We introduce VIALS, a visual question-answering benchmark with 161 such interpretation tasks, spanning the types of artifacts examined throughout experimental workflows in the biotech industry (rather than polished figures from publications and textbooks). While frontier vision-language models can now fluently describe natural images, we find that they are unable to accurately interpret these scientific images, reflecting limitations in domain knowledge and domain-specific visual reasoning capabilities. In contrast, scientists with relevant domain expertise find these visual interpretation tasks straightforward. AI that cannot similarly interpret such images will have limited utility in professional life sciences workflows, where such artifacts are central to how scientists reason, communicate, and make decisions.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-3b4b8a3ea9d72f81", "title": "AI with Authority, from Application to Silicon", "authors": ["Jason Hickey"], "year": 2026, "published": "2026-08-21T17:59:16Z", "updated": "2026-08-21T17:59:16Z", "arxiv_id": "http://arxiv.org/abs/2608.21356v1", "url": "https://arxiv.org/abs/2608.21356v1", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.AI", "cs.AR", "cs.LO"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["We publish the complete accounting: theorem provenance, a pre-registered token meter, floor-bounded human time, and an error ledger whose catch numbering runs to #256 --- a monotone counter over the mathematics campaign's append-only flags ledger, maintained 2026-07-07 to 2026-07-20 (one number, #79, was never assigned; later catches are recorded un-numbered) --- against zero incorrect proofs reaching the record."], "abstract": "For sixty years, machine verification has been a major cost overhead, affordable only for exceptional artifacts. Here we report that generative AI inverts this relationship: at AI speed, machine verification is not only economical but essential to productivity --- it is the incorruptible referee that lets one person safely direct autonomous machine work at scale. In five weeks, one researcher on consumer AI subscriptions directed a small fleet of AI agents from application code, through a verified compiler and executive, to a RISC-V processor taped out on a community silicon shuttle; no proof passed through human review, and no RTL was written by a human. The working discipline --- the Salt method --- rests on a proof kernel no hallucinated proof can pass: mathematical claims travel between agents as kernel-checked artifacts, and human attention is reserved for statements, designs, and rulings. Verification is stated link by link, from the Lean 4 kernel to SAT-checked equivalence at the silicon boundary. We publish the complete accounting: theorem provenance, a pre-registered token meter, floor-bounded human time, and an error ledger whose catch numbering runs to #256 --- a monotone counter over the mathematics campaign's append-only flags ledger, maintained 2026-07-07 to 2026-07-20 (one number, #79, was never assigned; later catches are recorded un-numbered) --- against zero incorrect proofs reaching the record.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-4f7fc6b9bc02cbef", "title": "Anatomy-Informed Neural Networks: Encoding Anatomic Priors in Loss and Architecture, with an SE(3) Formulation of Guidewire-Induced Aortoiliac Deformation", "authors": ["David P. Stonko"], "year": 2026, "published": "2026-08-21T17:38:42Z", "updated": "2026-08-21T17:38:42Z", "arxiv_id": "http://arxiv.org/abs/2608.21332v1", "url": "https://arxiv.org/abs/2608.21332v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CV", "cs.RO"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["The kinematics, loss and projection are verified against known ground truth; the mechanics solver only against its own optimality conditions, and predicted displacement is not yet mesh-converged.", "Future work will transfer this in silico model to real CT scans and test whether it improves predictive accuracy and reduces the training data required."], "abstract": "Deep-learning models of anatomy can be numerically plausible yet anatomically impossible, and they generalize poorly when data are scarce. We introduce Anatomy-Informed Neural Networks (AINN), in which soft anatomic priors enter as penalty terms in the loss (e.g., a branching penalty that treats a renal transplant artery off the iliac instead of the aorta as unexpected rather than impossible), in direct analogy to a physics-informed neural network, and hard anatomic priors (e.g., continuity of the vessel) are built into the architecture and state representation, making such invalid predictions impossible by construction wherever the prior admits architectural enforcement.\n  We develop it on a clinical test case with limited data: how the aortoiliac tree deforms when a stiff wire is introduced endoluminally. This is important to contemporary aortic surgery and will matter to autonomous endovascular navigation. We lift the vessel centerline and the wire path from R^3 to curves of frames in the Lie group SE(3), and couple a Cosserat-rod wire to a tortuosity-modulated, anatomically anchored vessel through a unilateral lumen-contact inequality. The prediction is a constrained minimizer of the coupled elastic energy, with contact forces as its Lagrange multipliers. Supervision is a Wasserstein-2 optimal-transport loss between the predicted projection through the C-arm geometry and the observed angiogram, so a 2D angiogram can train a 3D prediction. The kinematics, loss and projection are verified against known ground truth; the mechanics solver only against its own optimality conditions, and predicted displacement is not yet mesh-converged. Here, no network is trained. Future work will transfer this in silico model to real CT scans and test whether it improves predictive accuracy and reduces the training data required.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-0234b9dacab9939a", "title": "Re$^3$Cap: Retrieval-Guided Refinement for Image Captioning Enhancement via Reinforcement Learning", "authors": ["Haonan Jia", "Shichao Dong", "Zenghui Sun", "Jiawen Zheng", "Ziqi Miao", "Gege Shi", "Qiuyu Zhao", "Jinsong Lan", "Xiaoyong Zhu", "Bo Zheng"], "year": 2026, "published": "2026-08-21T17:07:41Z", "updated": "2026-08-21T17:07:41Z", "arxiv_id": "http://arxiv.org/abs/2608.21305v1", "url": "https://arxiv.org/abs/2608.21305v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Preference Optimization", "VLM"], "benchmarks": ["COCO"], "models": [], "key_results": ["Reinforcement Learning (RL) has demonstrated significant gains in image captioning, yet it is still limited in encouraging Large Vision-Language Models (LVLMs) to explore novel reasoning strategies.", "Especially, Re$^3$Cap outperforms GRPO with an average improvement of 8.64% in relation reasoning on the COCO-LN500 benchmark."], "abstract": "Reinforcement Learning (RL) has demonstrated significant gains in image captioning, yet it is still limited in encouraging Large Vision-Language Models (LVLMs) to explore novel reasoning strategies. This limitation leads to a performance gap between RL and Supervised Fine-Tuning (SFT). In this paper, we argue that multi-modal retrieval can serve as an effective reasoning signal for caption refinement. Based on this insight, we present the Retrieval-Guided Refinement for Image Captioning (Re$^3$Cap), a retrieval-guided reasoning strategy that enhances image captioning without requiring additional annotations. Instantiated by Caption Refinement Suggester (CRS) and Caption Quality Assessor (CQA), this strategy identifies hallucinations and omissions in image captions, leading to more accurate and detailed descriptions. Extensive experiments demonstrate the superiority of our method in image captioning, even compared with Supervised Fine-Tuning. Especially, Re$^3$Cap outperforms GRPO with an average improvement of 8.64% in relation reasoning on the COCO-LN500 benchmark.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-fa94bca6f6f80d2a", "title": "AUSO: Action-Level Unified Skill Optimization from Internalization to Utilization", "authors": ["Huizu Lin", "Chengkai Huang", "Tianqi Gao", "Tao Huang", "Daijiao Liu", "Tongxin Li", "Xiaoyan Sun", "Lina Yao"], "year": 2026, "published": "2026-08-21T16:53:27Z", "updated": "2026-08-21T16:53:27Z", "arxiv_id": "http://arxiv.org/abs/2608.21292v1", "url": "https://arxiv.org/abs/2608.21292v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["Skills play different roles as an agent's policy evolves: they should first provide learnable knowledge, then support capability formation, and finally be invoked only when they improve individual decisions.", "Experiments on ALFWorld, WebShop, and SearchQA show that AUSO consistently improves agent performance and out-of-distribution generalization over competitive baselines."], "abstract": "Skills play different roles as an agent's policy evolves: they should first provide learnable knowledge, then support capability formation, and finally be invoked only when they improve individual decisions. Existing methods rarely model this lifecycle. They either keep skills outside the model, fully internalize them, or select among internalization and utilization objectives through noisy task-level success rates. Such designs fragment training and assign uniform importance to actions within the same trajectory, even though skill guidance may help some decisions while distracting others. To solve these problems, we introduce AUSO (Action-level Unified Skill Optimization), which unifies skill learning and skill use through a progressive, action-aware optimization process. At the beginning of training, AUSO jointly learns from teacher guidance and environmental outcomes, enabling the policy to acquire foundational skills without losing task-oriented feedback. It subsequently emphasizes outcome-based policy optimization to consolidate autonomous problem-solving ability. As the policy matures, AUSO evaluates each sampled action under both skill-conditioned and skill-free contexts. The resulting action-level information signal is coupled with the trajectory outcome advantage, allowing beneficial skill-sensitive actions to receive stronger updates and harmful ones to be suppressed. Therefore, skills gradually transition from an external source of supervision into decision knowledge whose utilization is adapted to its action-level benefit, while reinforcement learning remains the shared backbone across all stages. Experiments on ALFWorld, WebShop, and SearchQA show that AUSO consistently improves agent performance and out-of-distribution generalization over competitive baselines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-46fa0ac316df17d7", "title": "Memory Augmentation Unlocks Efficient Chain-of-Thought Reasoning", "authors": ["Simeng Zhang", "Yilong Chen", "Wenyuan Zhang", "Zhenyu Zhang", "Yao Chen", "Junyuan Shang", "Tingwen Liu"], "year": 2026, "published": "2026-08-21T16:22:36Z", "updated": "2026-08-21T16:22:36Z", "arxiv_id": "http://arxiv.org/abs/2608.21265v1", "url": "https://arxiv.org/abs/2608.21265v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Long Context", "Memory / RAG"], "benchmarks": ["MMLU", "GSM8K", "MATH", "BBH"], "models": ["yielding"], "key_results": ["Experiments show that Memory consistently improves prompt-based Chain-of-Draft (CoD) compression across mathematical reasoning, complex reasoning, and science question answering tasks, yielding accuracy gains of 21.4, 28.0, 29.5, and 6.61 points over CoD on GSM8K, MATH, BBH, and MMLU-Sci, while achieving a 1.14--1.49$\\times$ latency speedup over standard CoT.", "Further analyzes show that the gains come from relevant reasoning memories rather than simply increasing context length."], "abstract": "Large language models often rely on Chain-of-Thought (CoT) reasoning to solve complex tasks, but verbose reasoning traces introduce substantial inference overhead. CoT compression shortens generation, yet aggressive compression may disrupt logical coherence and degrade performance. We formalize this trade-off as the \\textit{Context-Generation Substitution Law}, where explicit reasoning context substitutes for part of decode-time generation. Based on this principle, we propose \\textit{Memory-Augmented Compression}, a training-free framework that constructs reusable reasoning memories from historical traces and retrieves them as prefill-side scaffolds. Rather than using raw demonstrations, these memories summarize reusable reasoning patterns, key constraints, and critical operations to compensate for information lost during compression. Experiments show that Memory consistently improves prompt-based Chain-of-Draft (CoD) compression across mathematical reasoning, complex reasoning, and science question answering tasks, yielding accuracy gains of 21.4, 28.0, 29.5, and 6.61 points over CoD on GSM8K, MATH, BBH, and MMLU-Sci, while achieving a 1.14--1.49$\\times$ latency speedup over standard CoT. Memory is also compatible with token-level, reasoning-trace-level, and inference-state compression mechanisms. Further analyzes show that the gains come from relevant reasoning memories rather than simply increasing context length.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-207e11b008c52ef4", "title": "The Exceedance Design Effect: Effective Sample Size for Thresholds under Clustering", "authors": ["Adam Noonan"], "year": 2026, "published": "2026-08-21T16:19:50Z", "updated": "2026-08-21T16:19:50Z", "arxiv_id": "http://arxiv.org/abs/2608.21262v1", "url": "https://arxiv.org/abs/2608.21262v1", "primary_category": "stat.ML", "categories": ["stat.ML", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Chain-of-Thought", "Safety / Jailbreak", "Deep Research"], "benchmarks": [], "models": [], "key_results": ["Many machine-learning systems set a threshold at a quantile of a calibration set: conformal predictors that promise 90% coverage by drawing their cutoff at the calibration set's 90th percentile, abstention gates that decline to answer when a model's score falls below the calibration set's tenth percentile, safety filters that block any output scoring above the 99th percentile of a reference set."], "abstract": "Many machine-learning systems set a threshold at a quantile of a calibration set: conformal predictors that promise 90% coverage by drawing their cutoff at the calibration set's 90th percentile, abstention gates that decline to answer when a model's score falls below the calibration set's tenth percentile, safety filters that block any output scoring above the 99th percentile of a reference set. All of them promise that the threshold will hold at the stated rate on new data. The promise assumes the calibration examples are independent, and in modern pipelines they usually are not: they share a prompt, a document, a reasoning trace. Survey statistics has known how to discount correlated data since 1965, by counting how many independent observations a sample is worth, but only for averages. We show that a threshold needs a different count. The count depends on how often clustered scores land on the same side of the threshold, and that changes with where the threshold is set. How similar the scores are as numbers does not enter. We prove a closed-form law for the resulting effective sample size and for the spread of the coverage a deployed system actually sees.\n  Three consequences follow. The correction now used in the conformal literature is the wrong quantity, and can miss in either direction. A dataset has no single effective sample size. It has one for each level the threshold is set at. And the damage is invisible in coverage averaged over many runs, and fully felt by whoever deploys once. On a released calibration set of 25,028 examples, we measure the reliability of about 1,300.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-72d98aad352c38ac", "title": "EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering", "authors": ["Xuanyu Meng", "Jiashuo Sun", "Jash Rajesh Parekh", "Jiawei Han"], "year": 2026, "published": "2026-08-21T16:05:00Z", "updated": "2026-08-21T16:05:00Z", "arxiv_id": "http://arxiv.org/abs/2608.21252v1", "url": "https://arxiv.org/abs/2608.21252v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.DB", "cs.IR"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Across Loong and Oolong, EnSI-RAG achieves an average accuracy of 78.24."], "abstract": "Question answering (QA) over long, connected documents remains challenging because relevant evidence may span multiple entities and their relationships. Existing retrieval-augmented generation (RAG) methods typically index documents as raw chunks and retrieve them through embedding similarity. Their performance degrades when chunk boundaries separate entities from supporting evidence or when a question requires multi-hop reasoning across the corpus. We propose EnSI-RAG (Entity-Structure-Indexed Retrieval-Augmented Generation), a framework that constructs a query-independent, entity-centered index. Each record (e, t, k, v) represents an entity e, its type t, a semantic category k in {property, relation, aspect}, and a value v, while retaining links to the original source passages. At query time, these records serve as retrieval handles, and an LLM synthesizes the retrieved passages into the final answer. This design separates evidence localization from answer synthesis while preserving traceable source evidence. Across Loong and Oolong, EnSI-RAG achieves an average accuracy of 78.24. Relative to the published baseline scores used as references, this is 6.62 points higher, suggesting its effectiveness across these settings. The code is available at https://github.com/RamonMeng/EnSI-RAG.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-7a61e2eb50567a45", "title": "RARE: Decoupling Representation Steering from Expert Routing in Mixture-of-Experts Language Models", "authors": ["Zhibo Zhang", "Zhen Ouyang", "Ling Shi", "Kailong Wang"], "year": 2026, "published": "2026-08-21T15:45:36Z", "updated": "2026-08-21T15:45:36Z", "arxiv_id": "http://arxiv.org/abs/2608.21236v1", "url": "https://arxiv.org/abs/2608.21236v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["MoE", "Safety / Jailbreak"], "benchmarks": ["MMLU", "TruthfulQA"], "models": ["ying", "yielding"], "key_results": ["It further improves average TruthfulQA MC1 accuracy from 41.0% to 58.6% and CounterFact efficacy from 16.8% to 96.3%."], "abstract": "Representation engineering offers a lightweight means of controlling language-model behavior by modifying intermediate hidden states, but its direct application to Mixture-of-Experts (MoE) models introduces a structural mismatch. We first verify this failure mode through a series of empirical studies and find that preserving clean routing substantially recovers steering performance and that routing is more sensitive to semantic content than to behavioral changes under controlled content. Motivated by these findings, we introduce RARE, a router-agnostic representation engineering framework for MoE language models. RARE projects arbitrary behavioral perturbations onto the null space of the router matrix, thereby removing router-visible components, and further corrects routing drift propagated to selected downstream layers. To decide the best perturbation estimator in this framework, we evaluate five estimators on six heterogeneous open-weight MoE models across three steering scenarios: harmfulness, truthfulness, and factual editing. On harmfulness steering, RARE reaches an average attack success rate of 53.3% while retaining 67.8% MMLU accuracy, yielding a stronger aggregate effectiveness--utility trade-off than baselines. It further improves average TruthfulQA MC1 accuracy from 41.0% to 58.6% and CounterFact efficacy from 16.8% to 96.3%. These results support routing consistency as an important architectural consideration for adapting representation engineering to MoE models.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-c86296cbfebacac9", "title": "Enhancing LLMs in Predictive Political QA with Semi-Structured Data", "authors": ["Yinan Liu", "Zihan Zhou", "Zichun Jin", "Xinyu Wang", "Bin Wang", "Xiaochun Yang"], "year": 2026, "published": "2026-08-21T15:27:28Z", "updated": "2026-08-21T15:27:28Z", "arxiv_id": "http://arxiv.org/abs/2608.21218v1", "url": "https://arxiv.org/abs/2608.21218v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.IR"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": [], "key_results": ["Existing LLM augmentation methods, including actor-profile-based simulation and knowledge graph evidence injection, improve political reasoning but largely treat external resources as knowledge-based evidence, leaving prediction-relevant signals under-modeled.", "Across three real-world datasets and multiple LLMs, PSL consistently outperforms baselines, with ablations confirming the complementary gains of stance and structure signals."], "abstract": "Predictive political question answering (QA), such as predicting how a political actor will vote, goes beyond factual lookup. External political resources offer rich historical evidence, but rarely contain the answer itself. Existing LLM augmentation methods, including actor-profile-based simulation and knowledge graph evidence injection, improve political reasoning but largely treat external resources as knowledge-based evidence, leaving prediction-relevant signals under-modeled. We identify two complementary signals for predictive political QA: actor stances that capture issue-specific preferences, and high-order structure signals that capture indirect dependencies among political actors. We propose PSL, a dual-view framework that converts semi-structured political records into inference-oriented evidence for LLMs. PSL extracts stance signals from question-relevant actor records in a semantic view, and learns structure-aware actor representations from an actor interaction graph in a vector view. Across three real-world datasets and multiple LLMs, PSL consistently outperforms baselines, with ablations confirming the complementary gains of stance and structure signals.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-f22f0c8c69e82ddc", "title": "Personalized Privacy Control in LLMs via Attention Head Intervention", "authors": ["Junseok Kim", "Nakyeong Yang", "Kyomin Jung"], "year": 2026, "published": "2026-08-21T15:22:20Z", "updated": "2026-08-21T15:22:20Z", "arxiv_id": "http://arxiv.org/abs/2608.21209v1", "url": "https://arxiv.org/abs/2608.21209v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Test-time Scaling", "Multi-Agent"], "benchmarks": [], "models": ["Qwen2.5-7B", "Gemma3-4B showing average policy ignorance ratios of 51.25"], "key_results": ["Our method significantly improves adherence to user-specific privacy preferences by reducing cases where the model fails to follow the given policy."], "abstract": "The rise of agentic AI enables LLMs to access diverse user data, raising critical privacy concerns. Prior work on contextual privacy studies whether LLMs regulate information disclosure according to context-dependent norms. However, acceptable disclosure boundaries may vary across users even within the same context. To address this limitation, we introduce \\textit{personalized privacy}, which incorporates user-specific disclosure preferences into privacy control. We further present P3Bench~(\\textbf{P}ersonalized \\textbf{P}rivacy \\textbf{P}reservation \\textbf{Bench}mark), a novel benchmark extending contextual privacy policies with personalized disclosure policies. Experiments show that prompt-based policies fail to reliably enforce personalized privacy policies, with Qwen2.5-7B and Gemma3-4B showing average policy ignorance ratios of 51.25\\% and 74.28\\%, respectively. Finally, to address this problem, we propose \\textsc{Repair}, a robust inference-time attention head intervention method that adjusts disclosure behavior toward policy-consistent responses. Our method significantly improves adherence to user-specific privacy preferences by reducing cases where the model fails to follow the given policy.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-3f654a24a93c377a", "title": "Beyond Imitation: Self-Improving Robot Policies via Off-Policy Q-Planning", "authors": ["Varun Giridhar", "Anant Khandelwal", "Jeremy A. Collins", "Ignat Georgiev", "Animesh Garg"], "year": 2026, "published": "2026-08-21T15:18:37Z", "updated": "2026-08-21T15:18:37Z", "arxiv_id": "http://arxiv.org/abs/2608.21204v1", "url": "https://arxiv.org/abs/2608.21204v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Search / MCTS", "Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Behaviour Cloning (BC) has driven remarkable progress in robot manipulation, yet it is fundamentally limited by its inability to self-improve: a policy that fails cannot learn from that failure without additional human demonstrations.", "Reinforcement Learning fine-tuning offers a path to self-improvement but has proven difficult to scale to the multi-billion-parameter models underpinning modern robot policies.", "We exploit this asymmetry to enable value-guided action selection at inference (a single-step Q-weighted average over BC draws) and online self-improvement that fine-tunes only the Q-function, leaving the BC weights untouched."], "abstract": "Behaviour Cloning (BC) has driven remarkable progress in robot manipulation, yet it is fundamentally limited by its inability to self-improve: a policy that fails cannot learn from that failure without additional human demonstrations. Reinforcement Learning fine-tuning offers a path to self-improvement but has proven difficult to scale to the multi-billion-parameter models underpinning modern robot policies. We propose Q-Planning, which equips a large visuomotor BC policy with a small off-policy Q-function. Because a Q-function estimates value rather than imitates actions, it can be trained on the same successful demonstrations as the BC policy and later absorb both successful and failed deployment rollouts, an asymmetry BC does not have. We exploit this asymmetry to enable value-guided action selection at inference (a single-step Q-weighted average over BC draws) and online self-improvement that fine-tunes only the Q-function, leaving the BC weights untouched. On LIBERO and bimanual RoboTwin, ten iterations of self-improvement lift every benchmark score we tested (LIBERO-10 93% to 99%, RoboTwin 83.8% to 91.4%) and shorten successful episodes on the near-ceiling suites (LIBERO-Object, LIBERO-Goal). On two contact-rich bimanual real-robot tasks, the same loop (BC frozen, no human intervention) improves purely from its own deployment rollouts: stack-cups 40% to 90% and insert-wallet 25% to 80% in five iterations, whereas SFT on successful rollouts alone stalls at 55% and 30%. Under an identical online budget Q-Planning is the only method, among Best-of-N, filtered SFT, IBRL, DSRL, and DAWR, that improves stably from failures without training an auxiliary actor.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-9de412e673eed075", "title": "Tydra: An Efficient Hybrid Model for Tabular Data", "authors": ["Mieszko Komisarczyk", "Saurabh Mathur", "Maurice Kraus", "Sriraam Natarajan", "Kristian Kersting"], "year": 2026, "published": "2026-08-21T15:15:02Z", "updated": "2026-08-21T15:15:02Z", "arxiv_id": "http://arxiv.org/abs/2608.21199v1", "url": "https://arxiv.org/abs/2608.21199v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["Long Context"], "benchmarks": [], "models": [], "key_results": ["Transformer-based tabular foundation models such as TabPFN achieve strong predictive performance but incur quadratic computational cost with context length.", "Across 30 OpenML datasets, Tydra reduces inference time by 30% relative to TabPFN while retaining much of its predictive performance.", "Tydra also outperforms an approximately ten-times-larger Hydra model while providing faster inference."], "abstract": "Transformer-based tabular foundation models such as TabPFN achieve strong predictive performance but incur quadratic computational cost with context length. On the other hand, subquadratic SSM-based alternatives such as Hydra trade away accuracy for efficiency. To balance both, we introduce Tydra, a hybrid Transformer-State Space Model (SSM) architecture for tabular in-context learning that interleaves attention and SSM layers. Across 30 OpenML datasets, Tydra reduces inference time by 30% relative to TabPFN while retaining much of its predictive performance. Tydra also outperforms an approximately ten-times-larger Hydra model while providing faster inference. The results indicate that hybrid architectures are a promising direction for tabular foundation models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-12dc350102421fed", "title": "Towards Investigating Residual Hearing Loss: Quantification of Fibrosis in a Novel Cochlear OCT Dataset", "authors": ["Julia Dietlmeier", "Benjamin Greenberg", "Wenxuan He", "Teresa Wilson", "Rubing Xing", "Jordan Hill", "Adrienne Fettig", "Madeline Otto", "Teyhana Rounsavill", "Lina A. J. Reiss", "Jingang Yi", "Noel E. O'Connor", "George W. S. Burwood"], "year": 2026, "published": "2026-08-21T14:59:19Z", "updated": "2026-08-21T14:59:19Z", "arxiv_id": "http://arxiv.org/abs/2608.21189v1", "url": "https://arxiv.org/abs/2608.21189v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": [], "benchmarks": [], "models": ["ying"], "key_results": ["Results: We present the results of several state-of-the-art semantic segmentation models and compare their efficacy for identifying cochlear fibrosis and other relevant annotations, using a new library of manually segmented OCT images.", "Conclusions: We find that the best performance is achieved by using a modified version of the well-known UNET architecture (which we term 2D-OCT-UNET) that operates on the upscaled OCT input resolution."], "abstract": "Objective: Cochlear implants (CIs) are bionic prostheses that restores hearing via electrical stimulation of the auditory nerve. Hybrid CIs, which use electroacoustic stimulation (EAS), combine residual low-frequency acoustic hearing with CI electrical stimulation. Intracochlear fibrosis, which forms in response to the presence of the implant, may impede residual hearing function and gradually reduce the efficacy of EAS. It is therefore a translational objective to study the formation of cochlear fibrosis in rodents, with the goal of reducing fibrotic burden and improving outcomes for CI patients. Methods: We generate and annotate a novel dataset of optical coherence tomography (OCT) images from chronically implanted guinea pigs as part of an ongoing study focused on implant induced fibrosis. Objectively assessing fibrotic burden in this model, with high resolution and repeatability, presents an obvious use case for computer vision methods. Results: We present the results of several state-of-the-art semantic segmentation models and compare their efficacy for identifying cochlear fibrosis and other relevant annotations, using a new library of manually segmented OCT images. Conclusions: We find that the best performance is achieved by using a modified version of the well-known UNET architecture (which we term 2D-OCT-UNET) that operates on the upscaled OCT input resolution. Significance: For the first time, we have successfully applied computer vision techniques to an OCT dataset of implanted cochleae with fibrosis. Using this deep learning model, the cochlear fibrotic burden calculation can be reliably carried out as we verify in our experimental section. The dataset and the project code are available at: https://github.com/juliadietlmeier/CF-OCT-segmentation", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-a8044e9b5a8ba46e", "title": "SRL-MPC: Shape-Aware Reinforcement Learned Model Predictive Control", "authors": ["Ruihua Han", "Rui Gao", "Zhe Liu", "Xinyi Wang", "Chang Chen", "Shuai Wang", "Qi Hao", "Jia Pan", "Hengshuang Zhao"], "year": 2026, "published": "2026-08-21T14:46:06Z", "updated": "2026-08-21T14:46:06Z", "arxiv_id": "http://arxiv.org/abs/2608.21175v1", "url": "https://arxiv.org/abs/2608.21175v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.AI"], "pillar": "LLM Reasoning / Test-time Compute", "methods": ["RLVR / GRPO", "Safety / Jailbreak"], "benchmarks": [], "models": [], "key_results": ["The results show that SRL-MPC substantially outperforms representative baselines in safety and adaptability."], "abstract": "Safe and efficient shape-aware navigation in heterogeneous crowds and robot fleets remains challenging. Traditional approaches often assume homogeneous robots, sparse workspaces, simplified geometry, offline computation, or handcrafted parameters to make the problem tractable, which limits their deployment in dense crowd scenarios. Toward this end, we propose Shape-Aware Reinforcement Learned Model Predictive Control (SRL-MPC), a method for safe, efficient, and adaptive navigation in crowds with heterogeneous shapes without geometry simplification. To encode shape-aware safety, we formulate high-order control barrier function (HOCBF) constraints from geometric separation features (GSFs) based on support function transformation. A reinforcement learning (RL) framework then learns a neural policy that reads GSFs and outputs real-time MPC parameter updates, enabling the MPC solver to adapt to neighboring crowd geometries. The key advantage of SRL-MPC is that it preserves the safety structure and generalizability of MPC while integrating the adaptability and intelligence of RL. Experiments in randomized crowd scenarios with arbitrary shaped robot fleets demonstrate the effectiveness, scalability, and robustness of SRL-MPC. The results show that SRL-MPC substantially outperforms representative baselines in safety and adaptability. Project website: https://hanruihua.github.io/srl_mpc_project/", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-df3411e11c9e7153", "title": "Benchmarking Patent Drafting from Inventor-Style Disclosures", "authors": ["Lekang Jiang", "Wenjun Sun", "Stephan Goetz"], "year": 2026, "published": "2026-08-21T16:00:52Z", "updated": "2026-08-21T16:00:52Z", "arxiv_id": "http://arxiv.org/abs/2608.21249v1", "url": "https://arxiv.org/abs/2608.21249v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": [], "key_results": ["While recent large language models (LLMs) have achieved promising results on individual patent drafting tasks, they fundamentally fail to investigate the core challenge of real-world patent drafting: generating a complete and legally coherent patent application directly from early-stage invention materials.", "Benchmark results reveal that current LLMs exhibit limitations in patent drafting, while Patent-MAF provides a strong baseline that consistently outperforms evaluated open-source models and remains competitive with large closed-source models."], "abstract": "While recent large language models (LLMs) have achieved promising results on individual patent drafting tasks, they fundamentally fail to investigate the core challenge of real-world patent drafting: generating a complete and legally coherent patent application directly from early-stage invention materials. Prior work predominantly assumes later-stage, highly structured, or already legalistic inputs. However, real patenting workflows begin with informal, de-legalized disclosures authored by inventors. To bridge the gap, we introduce Dis2Pat, a disclosure-to-patent dataset that reflects realistic patenting workflows by requiring the generation of complete patent applications directly from inventor-style, de-legalized disclosures. Given the inherent difficulty of long-form, legally constrained patent drafting and the strong privacy requirements, we further propose a strong baseline named Patent-MAF. It is a multi-agent framework for locally deployable patent drafting. Benchmark results reveal that current LLMs exhibit limitations in patent drafting, while Patent-MAF provides a strong baseline that consistently outperforms evaluated open-source models and remains competitive with large closed-source models.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-31abc59bb63d4519", "title": "Utility Under Attack: Agent Memory Poisoning and the Limits of Content Screening and Provenance Ranking", "authors": ["Arulnidhi Karunanidhi"], "year": 2026, "published": "2026-08-21T15:37:35Z", "updated": "2026-08-21T15:37:35Z", "arxiv_id": "http://arxiv.org/abs/2608.21230v1", "url": "https://arxiv.org/abs/2608.21230v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG", "Computer Use"], "benchmarks": [], "models": [], "key_results": ["Poisoning 1.2% of a LongMemEval corpus reduces accuracy from 0.850 to 0.300."], "abstract": "Persistent memory makes false information durable: once a false statement is stored, it can be retrieved into future sessions that match it. We measure the cost of this failure mode using plainly worded false assertions generated in a single pass, with no instruction, trigger, or retriever optimization. Poisoning 1.2% of a LongMemEval corpus reduces accuracy from 0.850 to 0.300. A four-stage write-time screening pipeline that reaches 0.832 recall on indirect prompt injection while flagging 1.5% of trigger-word-laden benign text rejects 0 of 360 poisoned memories. We argue this exposes a boundary of content-only screening: distinguishing a false assertion from a true one generally requires external grounding beyond the text itself.\n  We then evaluate provenance-weighted retrieval. The shipped weight is statistically indistinguishable from no defense (p=0.80), while a stronger weight recovers utility only by excluding untrusted content. In a mixed-provenance corpus where untrusted content is mostly benign, accuracy rises from 0.3167 to 0.7000; when the answer-bearing evidence itself arrives untrusted, evidence recall falls to zero and accuracy to 0.0417. Under the measured similarity regime, the additive provenance term has no usable setting: a weight strong enough to resist query-shaped poison is also strong enough to suppress legitimate untrusted evidence. We therefore argue for bounded occupancy constraints at retrieval rather than additive provenance penalties, and release the harnesses, corpora, and aggregate run reports.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-de96bdae36587c71", "title": "Specification Portability Across LLM Development Agents: Cross-Agent Compatibility in Specification-Driven Software Migration", "authors": ["Oleg Grynets", "Oleksii Ilchuk", "Dariia Zatulna", "Vasyl Lyashkevych"], "year": 2026, "published": "2026-08-21T15:21:40Z", "updated": "2026-08-21T15:21:40Z", "arxiv_id": "http://arxiv.org/abs/2608.21208v1", "url": "https://arxiv.org/abs/2608.21208v1", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.AI", "cs.LO"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent", "Memory / RAG"], "benchmarks": [], "models": ["Claude Code and Cursor included in the initial single-agent evaluation. Native and foreign specifications were assessed using Token F1", "Gemini", "Gemini directly consumed a Kiro-origin specification", "Gemini in the tested configuration", "Gemini and Copilot. The findings suggest that specifications in heterogeneous SDD workflows should not automatically be treated as agent-neutral artifacts and motivate explicit consideration of specification portability"], "key_results": ["Rewriting substantially improved Gemini in the tested configuration, compression did not provide a universal benefit, and retrieval-augmented ingestion was the only common strategy represented on the per-agent Pareto frontiers of both Gemini and Copilot."], "abstract": "This paper investigates cross-agent specification portability using Oracle-to-PostgreSQL migration as a controlled software transformation task. The study combines two experimental stages. First, a specification-first migration pipeline was evaluated on 1,006 PL/SQL files, of which 623 were successfully regenerated and 380 generated scripts executed successfully in PostgreSQL 16. Second, cross-agent experiments were conducted on a dataset of 1,802 Oracle scripts with corresponding PostgreSQL implementations using Amazon Kiro, Google Gemini, and GitHub Copilot, with Claude Code and Cursor included in the initial single-agent evaluation. Native and foreign specifications were assessed using Token F1, exact match, SQL syntax validity, AST exact match, AST mean similarity, and immediate runnability. The results show that specification size alone does not predict implementation quality and that cross-agent transfer can produce substantial agent-dependent degradation. The strongest replicated case occurred when Gemini directly consumed a Kiro-origin specification, producing a Token F1 of 0.035, SQL syntax validity of 2.33%, and AST mean similarity of 0.015. Rewriting substantially improved Gemini in the tested configuration, compression did not provide a universal benefit, and retrieval-augmented ingestion was the only common strategy represented on the per-agent Pareto frontiers of both Gemini and Copilot. The findings suggest that specifications in heterogeneous SDD workflows should not automatically be treated as agent-neutral artifacts and motivate explicit consideration of specification portability, agent-specific interpretation, and retrieval-based access in multi-agent software engineering.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-dbbf0907cb4efd44", "title": "AID-Guard: Stateful Authorization for Delegated Agent Effects", "authors": ["Yingzhe Tong", "Leyu Dai", "Songhui Guo"], "year": 2026, "published": "2026-08-21T14:31:29Z", "updated": "2026-08-21T14:31:29Z", "arxiv_id": "http://arxiv.org/abs/2608.21159v1", "url": "https://arxiv.org/abs/2608.21159v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": ["yields"], "key_results": ["Tool-using AI agents turn delegated tasks into provider effects, yet authorization often ends at admission while provider state, delivery, retry, and recovery evolve."], "abstract": "Tool-using AI agents turn delegated tasks into provider effects, yet authorization often ends at admission while provider state, delivery, retry, and recovery evolve. A request may change before commit, or response loss may cause a replacement to create a second effect from one approval. We present AID-Guard, a stateful authorization-to-effect closure protocol. It revalidates the approved request and provider state at commit, retains one reservation under ambiguity, and permits release or one successor only after a terminal result or certified no effect with a delivery fence. For supported provider contracts, one reservation yields at most one effect across retry and recovery. To our knowledge, it is the first evaluated agent-authorization protocol to unify these controls in one lifecycle.\n  We implement a Python/SQLite prototype. In a declared loopback MCP domain, 13 live mutations caused no unauthorized provider effects, three concurrent histories were linearizable, and evidence bundles supported public verification and replay. All 210 Stripe provider-contract trials matched predeclared outcomes. Across Stripe and Resend, 40 terminalize-successor schedules, 30 overlapping races, and 10 crash-recovery schedules completed without duplicate effects. Under complete proposer compromise, AID-Guard blocked 44/44 attacks and admitted 44/44 matched legitimate proposals. Its strict exact-manifest profile reduced benign utility by 35.4 to 43.8 percentage points; a typed frontier recovered 9-10 completions without observed unsafe effects. A composition study blocked 20/20 post-admission lifecycle attacks and preserved 8/8 valid or exact-retry executions. The results support authorization-to-effect binding under the evaluated effect-path inventory, provider contracts, and failure schedules.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-d9304c27ea555341", "title": "Graph Engineering in the Era of LLM Agents: From Individual Intelligence to System Intelligence", "authors": ["Yuyuan Feng", "Zhishang Xiang", "Chaobin Yang", "Qichao Ma", "Zerui Chen", "Yujing Zhang", "Ke Huang", "Chuanjie Wu", "Zhaoxu Liu", "Yili Wang", "Xin He", "Jiapu Wang", "Zijin Hong", "Hao Chen", "Yuanchen Bei", "Kun Wang", "Shengyuan Chen", "Ningyu Zhang", "Enyan Dai", "Linhao Luo", "Qingyi Pan", "Qi Wang", "Wenqi Fan", "Guangjing Wang", "Na Zou", "Yangqiu Song", "Xin Wang", "Zechao Li", "Xia Hu", "Qing Li", "Xiao Huang", "Zhihong Zhang", "Jinsong Su", "Qinggang Zhang", "Yi Chang"], "year": 2026, "published": "2026-08-21T14:27:57Z", "updated": "2026-08-21T14:27:57Z", "arxiv_id": "http://arxiv.org/abs/2608.21156v1", "url": "https://arxiv.org/abs/2608.21156v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.AI", "cs.ET"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["This evolution has produced paradigms including Prompt Engineering to elicit model capabilities, Context Engineering to manage information access, Harness Engineering to organize external tools and resources, and Loop Engineering to support continual reflection and self-improvement.", "Achieving it requires more than adding agents; it demands explicit structures to organize work, coordinate heterogeneous agents, and maintain evolving execution states."], "abstract": "LLMs have evolved from language generators to autonomous agents capable of complex, long-horizon tasks. This evolution has produced paradigms including Prompt Engineering to elicit model capabilities, Context Engineering to manage information access, Harness Engineering to organize external tools and resources, and Loop Engineering to support continual reflection and self-improvement. Yet as tasks grow more complex, individual intelligence faces a fundamental limit: many tasks require heterogeneous expertise, interdependent subtasks, parallel execution, independent verification, and persistent state, exceeding any single agent's organizational capacity. Augmenting one agent's capabilities or context cannot resolve this architectural mismatch; intelligence must instead be distributed across specialized agents and organized at the system level. We call this System Intelligence: an agent system's ability to organize and coordinate multiple intelligent components into a coherent, adaptive whole pursuing a shared objective. Achieving it requires more than adding agents; it demands explicit structures to organize work, coordinate heterogeneous agents, and maintain evolving execution states. We introduce Graph Engineering, an emerging paradigm for next-generation agent systems. Unlike prior paradigms that mainly optimize individual interactions or agent-level behavior, Graph Engineering constructs explicit, dynamic, evolving graph structures representing tasks, agents, and system states. These abstractions provide a unified foundation for organizing complex objectives, orchestrating heterogeneous agents, modeling system dynamics, and enabling scalable agent evolution. We systematically review the principles, methodologies, and applications of Graph Engineering for LLM agents. Related papers, open-source data, and projects are collected at https://github.com/DEEP-JLU/Awesome-Graph-Engineering.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-d06c58e5d1b22f35", "title": "A Modular Agent for Reliable and Auditable Spatial Relation Verification in CT Scans", "authors": ["Simon Vincent Abel", "Heiko Hillenhagen", "Michael Götz", "Timo Ropinski", "Ayhan Can Erdur", "Daniel Santak Wolf"], "year": 2026, "published": "2026-08-21T14:16:46Z", "updated": "2026-08-21T14:16:46Z", "arxiv_id": "http://arxiv.org/abs/2608.21140v1", "url": "https://arxiv.org/abs/2608.21140v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Interpretability", "VLM"], "benchmarks": [], "models": ["Qwen2-VL"], "key_results": ["We evaluate the approach on the held-out MIRP spatial QA benchmark and compare it against representative end-to-end VLM baselines.", "The best-performing hybrid configuration reaches 94.1% accuracy and 94.2% F1, outperforming direct Qwen2-VL prompting by 42.5 percentage points in accuracy, while preserving interpretable intermediate representations and auditable reasoning stages."], "abstract": "Reliable spatial understanding is an important prerequisite for future medical vision-language systems that aim to support radiological report generation and structured image understanding. While modern vision-language models (VLMs) show promising performance on many medical imaging tasks, recent evidence suggests they remain weak in controlled spatial reasoning and often fail to reliably ground spatial relations in image evidence. Given that radiological reasoning hinges on understanding the relative positions of anatomical structures and findings, this spatial weakness poses risks to diagnostic accuracy. We present a modular medical imaging agent for binary spatial relation verification in axial CT slices. Instead of directly predicting spatial answers end-to-end, the system decomposes the task into explicit stages: language parsing, anatomical localization, and deterministic geometric verification. Natural-language queries are converted into structured relation tuples, queried organs are localized with a YOLO-based detector, and the final spatial decision is computed from object centers using deterministic geometric rules. We evaluate the approach on the held-out MIRP spatial QA benchmark and compare it against representative end-to-end VLM baselines. The best-performing hybrid configuration reaches 94.1% accuracy and 94.2% F1, outperforming direct Qwen2-VL prompting by 42.5 percentage points in accuracy, while preserving interpretable intermediate representations and auditable reasoning stages. The results suggest that explicit modular spatial verification can serve as a promising building block for future report-oriented medical imaging agents.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-aa559bbb77d93be0", "title": "CIVA: Critic-Induced Value-Subspace Attacks on Visual World-Model Agents", "authors": ["Jiancheng Wang", "Mingli Zhu", "Tong Zhang", "Jiaqi Ruan", "Wei Wang", "Siyuan Liang", "Dacheng Tao"], "year": 2026, "published": "2026-08-21T13:58:56Z", "updated": "2026-08-21T13:58:56Z", "arxiv_id": "http://arxiv.org/abs/2608.21114v1", "url": "https://arxiv.org/abs/2608.21114v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["World Model"], "benchmarks": ["DROP"], "models": ["V3"], "key_results": ["Extensive experiments on DMC walker walk, Atari Pong, and Crafter show that CIVA consistently outperforms five recent methods; on DMC walker walk, it achieves the largest reward drop of 26.07\\% while keeping temporal variation low, with TempAbs of 0.646."], "abstract": "Visual world-model agents such as DreamerV3 act through a recurrent latent state rather than a single observation, which weakens frame-wise observation attacks and makes their perturbations vary sharply over time under a strict per-frame perturbation constraint. We study white-box, causal, online attacks on such agents and propose Critic-Induced Value-Subspace Attacks (\\textbf{CIVA}). Our key observation is that, along a rollout, critic-guided perturbations concentrate in a low-dimensional subspace induced by the victim's own critic. Based on this observation, CIVA first probes the frozen victim offline with critic-guided PGD and extracts a low-rank value-subspace by SVD. At test time, it optimizes only the subspace coefficients, smooths them with an exponential moving average (EMA), and maps them back to pixels. This design attacks value-sensitive recurrent dynamics while keeping the online optimization cheap and temporally coherent. Extensive experiments on DMC walker walk, Atari Pong, and Crafter show that CIVA consistently outperforms five recent methods; on DMC walker walk, it achieves the largest reward drop of 26.07\\% while keeping temporal variation low, with TempAbs of 0.646.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-e934ae894dac6c9b", "title": "Large Language Models at the Intersection of Software Engineering and Software Security:An Evidence-Centered Structured Survey and Research Agenda", "authors": ["Wei Lin", "Tao Zhou", "Zhaofei Xie", "Changgui Hong"], "year": 2026, "published": "2026-08-21T13:54:38Z", "updated": "2026-08-21T13:54:38Z", "arxiv_id": "http://arxiv.org/abs/2608.21107v1", "url": "https://arxiv.org/abs/2608.21107v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.SE"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["The review shows that execution feedback and repository access can substantially improve engineering task completion, but do not by themselves establish security; conversely, static-analysis labels or vulnerability-classification scores rarely establish deployable correctness."], "abstract": "Large Language Models (LLMs) are moving from code completion toward repository-scale agents that retrieve context, edit files, execute tools, and participate in security-sensitive workflows. The evidence for these systems, however, remains divided between software engineering evaluations centered on functional task completion and software security evaluations centered on vulnerability detection, secure generation, or exploit-oriented validation. This evidence-centered structured survey synthesizes representative work available through May 31, 2026 across software engineering tasks, software security tasks, adaptation mechanisms, artifact granularity, and evaluation design. In addition to a task taxonomy, we introduce an assurance framework that separates functional correctness, security, operational reliability, evidence provenance, and agent authority. The review shows that execution feedback and repository access can substantially improve engineering task completion, but do not by themselves establish security; conversely, static-analysis labels or vulnerability-classification scores rarely establish deployable correctness. We identify recurring validity threats--weak test oracles, duplicated and temporally leaked data, changing agent harnesses, proxy-only security checks, and under-reported budgets and human intervention--and derive a minimum reporting protocol for cross-study comparison. The resulting research agenda prioritizes jointly secure-and-functional benchmarks, repository-scale threat models, calibrated human oversight, longitudinal maintainability evidence, and reproducible agent evaluation. The central conclusion is that model capability should be judged as an assurance case supported by task-appropriate evidence, rather than by a single benchmark score.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-ce1389d4346167d8", "title": "ClawSentry: A Progressive Multi-Tier Security Monitor for Safeguarding Autonomous LLM Agents", "authors": ["Kai Wang", "Zeming Wei", "BiaoJie Zeng", "Chang Jin", "An Wang", "Xiaokun Luan", "Zhixiao Lin", "Jingjing Qu", "Xia Hu", "Xingcheng Xu"], "year": 2026, "published": "2026-08-21T13:47:51Z", "updated": "2026-08-21T13:47:51Z", "arxiv_id": "http://arxiv.org/abs/2608.21101v1", "url": "https://arxiv.org/abs/2608.21101v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Safety / Jailbreak", "Multi-Agent"], "benchmarks": [], "models": ["GPT-5.4", "Claude Code", "Gemini CLI without modifying agent internals. On SkillInject with Codex", "ying", "Kimi CLI"], "key_results": ["As large language model (LLM) agents move from conversation to executing code, reading local files, and orchestrating external tools, a single agent hijacked by a malicious third-party skill can cause data exfiltration, privilege escalation, or cascading compromise."], "abstract": "As large language model (LLM) agents move from conversation to executing code, reading local files, and orchestrating external tools, a single agent hijacked by a malicious third-party skill can cause data exfiltration, privilege escalation, or cascading compromise. We argue that agentic risk is progressive: it can enter at four loci of the agent control loop--skill admission, invocation-time intent, execution-time effect, and post-action consequence--while a denied dangerous objective can reappear across surface forms, tools, or turns; existing safeguards are typically local to one lifecycle boundary or one call. Guided by this threat model, we present ClawSentry, an open-source, framework-agnostic security supervision gateway for agent runtimes. Before a skill package is ever executed, First-use Skill Package Review (FSPR) audits it under a deterministic evidence floor, escalating unresolved cases to bounded read-only agentic review (locus A). At runtime, a three-tier progressive decision engine--a deterministic L1 layer, a rule-anchored L2 semantic reviewer, and a read-only L3 evidence-seeking agent--spends contextual review only on the residual ambiguity, while a session-level anti-bypass mechanism recognizes tool-switching and rephrased retries (loci B--C); a post-action path feeds high-severity evidence non-retroactively into later review (locus D). An Agent Harness Protocol (AHP) abstraction applies one policy across Codex, Claude Code, Kimi CLI, and Gemini CLI without modifying agent internals. On SkillInject with Codex/GPT-5.4, contextual ASR falls from 39.55% to 2.61% while contextual TSR moves only from 83.78% to 83.05%. Across five Work Agents on the full SkillsSafety benchmark, ClawSentry confines ASR to 9.09--15.03% from 33.5--49.7% unprotected, and aggregate TSR on clean skills remains 98.7%.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-0bc898c087e98e6c", "title": "ReFrame: Evidence-Guided Test-Time Safety Alignment in Multimodal Large Language Models", "authors": ["Wenzheng Jiang", "Xuankun Rong", "Yuanzhao Zhai", "Dawei Feng", "Huaimin Wang"], "year": 2026, "published": "2026-08-21T13:47:21Z", "updated": "2026-08-21T13:47:21Z", "arxiv_id": "http://arxiv.org/abs/2608.21100v1", "url": "https://arxiv.org/abs/2608.21100v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Test-time Scaling", "Preference Optimization", "Safety / Jailbreak", "VLM"], "benchmarks": [], "models": ["ying"], "key_results": ["Experiments across multiple MLLMs and benchmarks show that ReFrame improves jailbreak defense, safety awareness, and oversensitivity reduction while preserving multimodal utility."], "abstract": "While multimodal large language models (MLLMs) extend model capabilities beyond text, they also make safety alignment increasingly challenging. Multimodal safety alignment methods must address cross-modal jailbreaks, safety-awareness failures, and over-sensitive refusals. However, existing methods often rely on retraining or internal-state inspection, limiting their applicability to deployed closed-source MLLMs and motivating test-time safety alignment. We analyze this setting and identify two key obstacles, utility dominance and reasoning inertia, which cause models to overlook latent risks or follow malicious reasoning trajectories. Guided by these insights, we propose ReFrame, a training-free multimodal input reframing framework where two agents share a lightweight locally deployed MLLM: the evidence-generation agent constructs complementary risk and utility evidence, and the rewrite-and-routing agent converts it into a safe proxy prompt and image-routing decision before calling the downstream MLLM, without modifying it or accessing its internal information. Experiments across multiple MLLMs and benchmarks show that ReFrame improves jailbreak defense, safety awareness, and oversensitivity reduction while preserving multimodal utility.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-24ddbe91a622a3cd", "title": "Trustworthy RAG: An Evaluation Agent for Detecting Misinformation and Knowledge Poisoning in Generative AI Systems", "authors": ["Balkrishna Giri", "Md Toufique Hasan", "Jussi Rasku", "Muhammad Waseem", "Pekka Abrahamsson"], "year": 2026, "published": "2026-08-21T13:42:41Z", "updated": "2026-08-21T13:42:41Z", "arxiv_id": "http://arxiv.org/abs/2608.21095v1", "url": "https://arxiv.org/abs/2608.21095v1", "primary_category": "cs.SE", "categories": ["cs.SE", "cs.AI", "cs.CL", "cs.CR", "cs.IR"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Memory / RAG"], "benchmarks": ["TruthfulQA"], "models": ["Llama 3.3"], "key_results": ["Retrieval-Augmented Generation (RAG) grounds Large Language Model (LLM) outputs in external knowledge, but RAG systems usually trust whatever they retrieve, creating a Security-Reliability Gap: high semantic relevance does not guarantee factual truth."], "abstract": "Retrieval-Augmented Generation (RAG) grounds Large Language Model (LLM) outputs in external knowledge, but RAG systems usually trust whatever they retrieve, creating a Security-Reliability Gap: high semantic relevance does not guarantee factual truth. Adversaries exploit this through knowledge poisoning, inserting malicious documents to cause targeted misinformation. We propose an Evaluation Agent, middleware that combines Natural Language Inference (NLI) factual verification, a five-signal poison detector with relevance-weighted aggregation, and a Trust Index T = 0.4 F + 0.35 C + 0.25 (1 - P ) with a non-linear dampener for high-contamination contexts. On TruthfulQA with Llama 3.3 70B, the agent reaches 91% accuracy and 100% precision, with 100% recall on instruction injection, while in-place edits, such as entity swaps, remain hard to detect. Across three LLMs the Trust Index stays discriminative, with a Receiver Operating Characteristic Area Under the Curve (ROC-AUC) of 0.73 to 0.81; generation style matters more than model size, and per-LLM threshold calibration restores baseline competitive accuracy, whereas a weaker FEVER result shows that cross-dataset generalization requires domain-specific calibration. In a software-engineering use case, a secure-coding assistant over guidance from the Open Worldwide Application Security Project (OWASP) Top 10 and the Common Weakness Enumeration (CWE), the agent reliably blocks instruction injection of unsafe advice (F1 92%), while contradiction and subtle semantic weakening remain hard. Throughout, the agent measures detection of poisoned context before generation, not whether the LLM adopts the injected misinformation. We release the proposed approach, attack generator, and experimental artifacts at the link: https://github.com/GPT-Laboratory/TrustworthyRAG.", "code_available": true, "confidence": 0.9, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-bf352fbfa4ce2baf", "title": "Jokes Aside: Measuring the Semantic Distance of Double Meanings", "authors": ["Fabio De Ponte"], "year": 2026, "published": "2026-08-21T13:31:46Z", "updated": "2026-08-21T13:31:46Z", "arxiv_id": "http://arxiv.org/abs/2608.21087v1", "url": "https://arxiv.org/abs/2608.21087v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Agentic AI / Deep Research Systems", "methods": [], "benchmarks": [], "models": [], "key_results": ["Results revealed that models trained on the proposed metrics performed poorly in predicting humor ratings: on JokeJudger, the best model achieved 57.1% accuracy, below the 61.5% baseline, while performance on Expunations and rJokes was even lower."], "abstract": "Large language models have significantly enriched the toolkit for computational humor research, particularly in the automated generation of jokes and puns. A key innovation, contextual embedding vectors, offers new opportunities to revisit and refine earlier hypotheses. Notably, Petrovic and Matthews (2013) proposed a joke generation model based on the scheme \"I like my X like I like my Y, Z\" (e.g. \"I like my ice like I like my dreams, crushed\"). They suggested that joke hilarity increases with: a) frequent association of Z with X and Y, b) rarity of Z, c) ambiguity of Z, and d) meaning distance between X and Y. Building on this, Winters et al. (2019) proposed a set of metrics, based on Google Ngrams and Word2Vector. In this work, three out of their five metrics are revisited with word embeddings: obviousness, compatibility, and comparison. Another measure, symmetry, defined as closeness of Z to both X and Y, is introduced here for the first time. Two models were used to collect the embedding vectors (OpenAI text-embedding-3-small and MiniLM all-MiniLM-L6-v2) on three datasets: JokeJudger, Expunations, and rJokes. The last two datasets, Expunations, and rJokes, were expanded by adding paired sentences that captured the ambiguous expression at the core of each joke in its two different meanings. Results revealed that models trained on the proposed metrics performed poorly in predicting humor ratings: on JokeJudger, the best model achieved 57.1% accuracy, below the 61.5% baseline, while performance on Expunations and rJokes was even lower. Nevertheless, the symmetry metric seems consistently associated with higher-rated jokes, suggesting it may capture a necessary -though not sufficient- property of humor.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-5199fc40ab9c8ebc", "title": "CoAnchor: Robust Collaborative Perception under Spatio-Temporal Misalignment via Object-Level Anchors", "authors": ["Chi Li", "Rui Lin", "Aobo Ji", "Dongzhu Xu"], "year": 2026, "published": "2026-08-21T12:52:38Z", "updated": "2026-08-21T12:52:38Z", "arxiv_id": "http://arxiv.org/abs/2608.21055v1", "url": "https://arxiv.org/abs/2608.21055v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Preference Optimization"], "benchmarks": [], "models": [], "key_results": ["Collaborative perception extends the sensing range of a single vehicle by fusing observations from nearby agents, which improves the robustness of autonomous driving.", "Extensive experiments on both simulated and real-world datasets illustrate that CoAnchor remains competitive under clean settings and improves the robustness under joint delay and pose perturbations with a favorable practical accuracy-efficiency trade-off."], "abstract": "Collaborative perception extends the sensing range of a single vehicle by fusing observations from nearby agents, which improves the robustness of autonomous driving. In realistic deployments, however, the received collaborator messages are often affected by both communication delay and relative-pose noise, which jointly cause stale observations, spatial misalignment, and unstable feature fusion. Existing methods usually address these issues from either the spatial or temporal side, but handling them jointly in a unified and efficient manner remains challenging. In this paper, we propose CoAnchor, an anchor-centric spatio-temporal alignment framework for asynchronous collaborative perception. Instead of directly reasoning on dense BEV features, CoAnchor builds sparse object-level spatio-temporal anchors as a shared interface for pose correction and tightly connects spatial refinement, temporal propagation, and current-time verification within one unified loop, while keeping the overall correction process lightweight. Extensive experiments on both simulated and real-world datasets illustrate that CoAnchor remains competitive under clean settings and improves the robustness under joint delay and pose perturbations with a favorable practical accuracy-efficiency trade-off.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-8aabef2e371253c5", "title": "$Z^2$-ACT: End-to-End Verifiable Agentic Intent Control for Open 6G RAN", "authors": ["Sunder Ali Khowaja", "Kapal Dev", "George C. Alexandropoulos"], "year": 2026, "published": "2026-08-21T12:44:03Z", "updated": "2026-08-21T12:44:03Z", "arxiv_id": "http://arxiv.org/abs/2608.21049v1", "url": "https://arxiv.org/abs/2608.21049v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI", "cs.NI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Safety / Jailbreak", "Multi-Agent"], "benchmarks": [], "models": ["phic"], "key_results": ["Our experimental evaluation on public ColO-RAN measurements compares the full architecture against targeted ablations and a conventional reinforcement-learning baseline.", "Results indicate improved actuation filtering and attack resilience at modest latency and signaling cost inside the near-real-time envelope."], "abstract": "With the progression in open and disaggregated 6G radio access networks, it is expected that the system will be able to host multi-vendors. In order to host multi-vendors, it is essential that AI-assisted control loops remain safe, verifiable, and auditable under concurrent operator intents and untrusted model inputs. The existing studies address the agentic coordination, formal intent constraints, zero-trust prompt verification and cryptographic accountability in isolation, which leaves pre-realization safety, continuous semantic verification and cross-domain audit incomplete when used individually. In this regard, we propose zero-knowledge auditable control and zero-trust verifiable agentic intent architecture ($Z^2$-ACT), which integrates the aforementioned four primitives across the non-real-time and near-real-time RICs. We encode the typed Intent Contracts as operator goals while the large language model inputs are only admitted after a practical adversarial intent check. The skill sequences in the proposed study are released only when a self-management gate is satisfied while every successful commit is recorded as a binding commitment with a zero-knowledge proof. Our experimental evaluation on public ColO-RAN measurements compares the full architecture against targeted ablations and a conventional reinforcement-learning baseline. A live large language model is used in the non-real-time path to translate operator intents into Intent Contracts; we report translation accuracy, the rate of invalid or hallucinated contracts, non-real-time latency, and behavior under adversarial or misleading intents. Near-real-time control remains trace-driven on the public KPM sequences. Results indicate improved actuation filtering and attack resilience at modest latency and signaling cost inside the near-real-time envelope.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-6c15a783413ac3e2", "title": "Don't Solve, Just Compare: Tiny Advisors for Runtime Intervention in LLM Agents", "authors": ["Yanze Jiang", "Mingxuan Li", "Yuhao Wang", "Shengfang Zhai", "Jiaheng Zhang"], "year": 2026, "published": "2026-08-21T12:20:14Z", "updated": "2026-08-21T12:20:14Z", "arxiv_id": "http://arxiv.org/abs/2608.21027v1", "url": "https://arxiv.org/abs/2608.21027v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Tool Use"], "benchmarks": [], "models": ["ying"], "key_results": ["As these agents operate over longer horizons, runtime intervention offers a way to improve reliability without retraining the underlying actor.", "Across WebShop, ALFWorld, and tau^3-Retail with three actors, COTA improves all nine evaluation settings and outperforms the compared baselines."], "abstract": "LLM agents are emerging as an important paradigm for real-world tasks that require reasoning, tool use, and sequential decision-making. As these agents operate over longer horizons, runtime intervention offers a way to improve reliability without retraining the underlying actor. Failure detection alone is insufficient. Effective intervention must also provide a useful direction for recovery. Existing approaches often rely on an expert solver or a critic that generates task-specific corrections, incurring either the cost of another capable solver or the capacity demands of a task-capable critic. We introduce Comparison-Only Tiny Advisor (COTA), a comparison-only framework for constructive runtime intervention. In COTA, a tiny comparator judges whether sampled alternatives lead to better continuations than the actor's proposal, and repeated comparisons determine when intervention is warranted. We train the comparator using pairwise supervision constructed from same-prefix counterfactual branches. Preferred alternatives are returned as non-binding advice, leaving the original actor to replan. Across WebShop, ALFWorld, and tau^3-Retail with three actors, COTA improves all nine evaluation settings and outperforms the compared baselines. These results show that constructive runtime intervention can remain effective even when the auxiliary model has substantially weaker task-solving capability than the actor.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-4626e7c0eaaef25c", "title": "Belief Without Behavior: Measuring the Translation of Theory of Mind into Coordinated Social Action in Vision-Language Models", "authors": ["Tonglin Yan", "Gregoire Sergeant-Perthuis", "David Rudrauf"], "year": 2026, "published": "2026-08-21T10:55:14Z", "updated": "2026-08-21T10:55:14Z", "arxiv_id": "http://arxiv.org/abs/2608.20975v1", "url": "https://arxiv.org/abs/2608.20975v1", "primary_category": "cs.AI", "categories": ["cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Effective social interaction requires agents to translate mental state inferences into coordinated behavioral signals across verbal and nonverbal channels simultaneously."], "abstract": "Effective social interaction requires agents to translate mental state inferences into coordinated behavioral signals across verbal and nonverbal channels simultaneously. Yet existing benchmarks evaluate theory of mind (ToM) reasoning and embodied behavior in isolation, leaving unmeasured the gap between social inference and social action. We introduce MOSAIC (Multimodal Orchestration of Social Action, Inference, and Communication), a controlled benchmark in which two embodied agents interact across cooperative and competitive scenarios requiring integration of verbal statements, spatial trajectories, gaze direction, and facial expression under systematically varied ToM constraints. Evaluating 13 models, including 11 VLMs, across 200 trials per model, we find that VLMs fail to produce behaviors consistent with the expected outcomes under ToM-order constraints, and that imposing explicit ToM-order constraints produces no reliable behavioral change aligned with the specified reasoning level. Signal-level analysis reveals two sequential bottlenecks: most models cannot produce directionally coherent nonverbal signals, and even when signals are present, VLM agents fail to interpret others behaviors and react to them. PCM-LLM, included as a structured architectural reference point with an explicit ToM module, succeeds across all conditions, suggesting that explicit belief-action coupling is a sufficient ingredient for this class of tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-d4b1c876c0009b09", "title": "Vibe Coding and Web Application Security: A Twin-Prompt Study", "authors": ["Darko Andročec"], "year": 2026, "published": "2026-08-21T10:34:41Z", "updated": "2026-08-21T10:34:41Z", "arxiv_id": "http://arxiv.org/abs/2608.20963v1", "url": "https://arxiv.org/abs/2608.20963v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.AI"], "pillar": "Agentic AI / Deep Research Systems", "methods": ["Multi-Agent"], "benchmarks": [], "models": ["yielding"], "key_results": ["Large language models increasingly generate complete web applications from natural-language prompts, raising the question of whether explicitly requesting security best practice improves the result."], "abstract": "Large language models increasingly generate complete web applications from natural-language prompts, raising the question of whether explicitly requesting security best practice improves the result. We study six functionally distinct web applications, each generated in two prompt variants that are identical except for an appended security-requirements section: a baseline (A) and a security-aware (B) variant. All twelve programs were produced by the same agentic coding assistant and the same model version in a single, non-iterative generation round, and were then analyzed with static, dependency, dynamic and manual techniques, yielding 75 confirmed findings out of 85 candidates. The security-aware variant produced fewer confirmed findings in every application (24 versus 51) and contained no Critical or High issues; the most severe finding was detected only by manual testing. Because the corpus is small and each variant was generated once, we report descriptive observations rather than statistically established effects, and position the work as a preliminary study whose pipeline is being scaled to multiple models and repeated runs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-b3ba2bda56acb68a", "title": "Rethinking Expressivity and Efficiency in Test-Time Training", "authors": ["Zeyun Zhong", "Joya Chen", "Manuel Martin", "Frederik Diederichs", "Juergen Gall", "Juergen Beyerer"], "year": 2026, "published": "2026-08-21T17:12:28Z", "updated": "2026-08-21T17:12:28Z", "arxiv_id": "http://arxiv.org/abs/2608.21308v1", "url": "https://arxiv.org/abs/2608.21308v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Test-time Scaling", "Long Context"], "benchmarks": [], "models": [], "key_results": ["It performs on par with previous TTT and hybrid attention baselines in language modeling while outperforming them on in-context retrieval."], "abstract": "Test-Time Training (TTT) enables long-context processing via continuous weight updates during inference, but current methods struggle to balance the expressivity of per-token update dynamics with the hardware efficiency of chunk-wise approximations. We propose E$^2$-TTT (Expressive and Efficient TTT) to bridge this gap. Under the standard approximation of taking gradients at the chunk-start weights, we derive a closed-form state transition that exactly reproduces the chunk-end fast-weight and momentum states of the per-token recurrence. This enables fully parallelized chunk-level training while preserving the temporal structure of the update rule that prior chunk-wise methods discard. We validate E$^2$-TTT by training models up to 1.3B parameters from scratch. It performs on par with previous TTT and hybrid attention baselines in language modeling while outperforming them on in-context retrieval. Its advantage is most pronounced in length extrapolation: on the standard ``Needle in a Haystack'' passkey test, it retains over 90% accuracy at $8\\times$ the training context length. Meanwhile, E$^2$-TTT can match the training throughput of efficient chunk-wise methods, demonstrating that it effectively reconciles expressivity with efficiency. The code is available at https://github.com/zeyun-zhong/E2-TTT.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-a458362bae0473fc", "title": "Llama-Mobile: Efficient 2.7-Bit Quantization of VLMs", "authors": ["Luka Ribar", "Jeevan Bhoot", "Douglas Orr"], "year": 2026, "published": "2026-08-21T14:10:31Z", "updated": "2026-08-21T14:10:31Z", "arxiv_id": "http://arxiv.org/abs/2608.21134v1", "url": "https://arxiv.org/abs/2608.21134v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization", "Memory / RAG", "VLM"], "benchmarks": [], "models": ["Llama-Mobile", "Llama 3.2", "ying"], "key_results": ["Deploying vision-language models (VLMs) on mobile devices is challenging due to their significant memory and compute requirements."], "abstract": "Deploying vision-language models (VLMs) on mobile devices is challenging due to their significant memory and compute requirements. We present a framework for quantizing VLMs for efficient inference on resource-constrained hardware. Our approach combines a quantization pipeline that uses the model itself to generate training data and does not require access to the training setup, with a novel 2.7-bit-per-parameter format supporting efficient execution on Arm CPUs. We validate our approach by compressing the Llama 3.2 11B Vision Instruct model to 3.7 GB with 8-bit activations, preserving strong performance on a set of standard visual question answering tasks.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-9f0561c0e17c5b96", "title": "COMET: Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models", "authors": ["Chenghua Zhu", "Zhaolu Kang", "Qifan Shi", "Siyan Wu", "Kehan Jiang", "Lei Wei", "Lianyu Hu", "Guangyuan Dong", "Mingbo Yang", "Rui Lu", "Guibo Luo"], "year": 2026, "published": "2026-08-21T12:28:36Z", "updated": "2026-08-21T12:28:36Z", "arxiv_id": "http://arxiv.org/abs/2608.21030v1", "url": "https://arxiv.org/abs/2608.21030v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.CL", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["RLVR / GRPO", "Distillation", "VLM"], "benchmarks": [], "models": ["Qwen3-VL-8B", "InternVL2.5-8B", "LLaVA-178K"], "key_results": ["The method achieves consistent overall improvements with a pronounced motion-temporal bias: on Qwen3-VL-8B, action-centric tasks (STAR, SSv2) improve by 4.9% on average, temporal reasoning tasks (NExT-QA, CLEVRER, LLaVA-178K) by 2.1% over BL-GRPO, while static perception tasks (PerceptionTest) remain on par."], "abstract": "Video multimodal large language models have advanced significantly, yet fine-grained motion-temporal understanding remains fragile. The core bottleneck is not only sparse frame sampling, but also the lack of a complete temporal modeling pipeline for explicitly representing frame-to-frame change, enabling appearance-motion interaction, and optimizing temporal direction sensitivity. We propose COMET, a temporally grounded framework that systematically strengthens video MLLMs through explicit temporal representation, appearance-motion fusion, and direction-aware optimization. Architecturally, COMET introduces a temporal motion branch built on Taylor frame differences and injects its motion evidence into the appearance stream via temporal attention bias-enhanced cross-attention. For optimization, COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage that turns temporal order into a direct learning signal and strengthens the model's use of directional motion patterns encoded by the temporal motion branch. The method achieves consistent overall improvements with a pronounced motion-temporal bias: on Qwen3-VL-8B, action-centric tasks (STAR, SSv2) improve by 4.9% on average, temporal reasoning tasks (NExT-QA, CLEVRER, LLaVA-178K) by 2.1% over BL-GRPO, while static perception tasks (PerceptionTest) remain on par. The same gain pattern also transfers to InternVL2.5-8B, indicating that COMET generalizes across model families.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-c4d28bc287f04c65", "title": "Target-Aware Calibration Data Selection for Preserving Uncertainty in Quantized Language Models", "authors": ["Zhen Yang", "Sizai Hou", "Kaiwen Zheng", "Yaofang Liu", "Liang He", "Yixuan Chen", "Kangning Cui"], "year": 2026, "published": "2026-08-21T12:07:49Z", "updated": "2026-08-21T12:07:49Z", "arxiv_id": "http://arxiv.org/abs/2608.21019v1", "url": "https://arxiv.org/abs/2608.21019v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["Quantization is widely used to deploy large language models, but its effect on uncertainty behavior, such as confidence, margins, and abstention, is rarely treated as a primary objective."], "abstract": "Quantization is widely used to deploy large language models, but its effect on uncertainty behavior, such as confidence, margins, and abstention, is rarely treated as a primary objective. We frame calibration-data selection for quantization as a target-dependent uncertainty-preservation problem. Different deployments emphasize different regions of the input distribution, yet prior work mainly optimizes accuracy-oriented compression metrics or adjusts scores after quantization. We formalize this goal with distributional and boundary preservation risks, and provide a simple mixture-mismatch argument explaining why no single calibration recipe should be expected to fit all targets. We introduce Doubt-Preserving Quantization (DPQ), a lightweight pre-quantization recipe family that uses full-precision predictions to construct target-aligned calibration mixtures of high-doubt examples and generic anchors. Across 8 language models, 9 NLP benchmarks, and 22 comparison methods, the leading fixed recipe changes with the preservation target: DPQ-r75 leads on SQuAD2 answerability-boundary preservation, while milder or single-signal variants, including DPQ-r50, confidence-only, and entropy-only, better preserve broad multiple-choice QA behavior. These results show that calibration data should be selected for the specific full-precision score behavior a deployment needs to preserve, rather than treated as a fixed quantization detail.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-5e588ff1f52c4048", "title": "From a Static Multi-Level Small Semantic Codebook to a Dynamic Single-Level Large Semantic Codebook for Generative Recommendation", "authors": ["Tianlu Xie", "Xin Ku", "Mingjie Sun", "Yunhao Sha", "Lixiang Wang", "Peng Wang", "Yiyu Wang", "Wenjin Wu", "Zhaojie Liu", "Peng Jiang", "Wenwu Ou"], "year": 2026, "published": "2026-08-21T11:58:44Z", "updated": "2026-08-21T11:58:44Z", "arxiv_id": "http://arxiv.org/abs/2608.21012v1", "url": "https://arxiv.org/abs/2608.21012v1", "primary_category": "cs.IR", "categories": ["cs.IR", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["On two public datasets, the two-level SID improves mean Recall@10 by 5.0%-8.8% and mean NDCG@10 by 4.1%-5.1% for OneRec-V1, and by 7.1%-8.7% and 3.8%-8.5%, respectively, for OneRec-V2.", "Dynamic updating provides further gains on KuaiRec.", "Across three serving architectures, the shorter SID reduces estimated autoregressive-decoding FLOPs by 47.93%-48.70% and increases single-card QPS by 28.57%-47.0%."], "abstract": "Generative recommendation represents each item with a sequence of discrete Semantic IDs (SIDs) and predicts the sequence to retrieve the next item. Typical systems use multi-level residual quantization, which increases autoregressive decoding cost and creates a large hierarchical space that may be sparsely occupied. Static codebooks also become misaligned with current traffic as new items arrive and exposure distributions change. We propose a single-level large semantic codebook that replaces multiple residual semantic codes with one semantic token while retaining a separate collaborative disambiguation token to reduce item collisions. We further introduce an exposure-aware dynamic update mechanism based on temporal weight decay, exponential moving-average center updates, and an exposure-weighted penalty on SID changes. We also develop an offline evaluation framework covering representation quality, code utilization, cluster load, full-SID collision, and temporal stability. On two public datasets, the two-level SID improves mean Recall@10 by 5.0%-8.8% and mean NDCG@10 by 4.1%-5.1% for OneRec-V1, and by 7.1%-8.7% and 3.8%-8.5%, respectively, for OneRec-V2. Dynamic updating provides further gains on KuaiRec. Across three serving architectures, the shorter SID reduces estimated autoregressive-decoding FLOPs by 47.93%-48.70% and increases single-card QPS by 28.57%-47.0%. A five-day online A/B test serving 2.5% of production traffic improves the primary consumption metric by 0.792%.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-db4d80a323657a75", "title": "Jacobian-guided Noise Injection for Quantization Robustness in Large Language Models", "authors": ["Deepanshu Pandey", "Arnav Chavan", "Nahush Lele", "Sankalp Dayal", "Deepak Gupta"], "year": 2026, "published": "2026-08-21T11:16:40Z", "updated": "2026-08-21T11:16:40Z", "arxiv_id": "http://arxiv.org/abs/2608.20988v1", "url": "https://arxiv.org/abs/2608.20988v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": ["ImageNet", "WikiText"], "models": [], "key_results": ["We evaluate the method on SOTA LLM architectures, where it demonstrates improved robustness over popular PTQ methods.", "Empirical analysis reveals that the proposed method gives up to +37% relative gains on Top-1 accuracy on ImageNet-1K for SigLIP and improves relative perplexity by upto 40% on WikiText for language models in low bit quantisation settings, proving the efficacy of the approach."], "abstract": "Quantization of Large Language Models (LLMs) is often hindered by the sensitivity of the self-attention mechanism to discretization errors. We identify the softmax operator as a bottleneck for quantization stability due to its sensitivity to outliers and state-dependent Jacobian. We theoretically establish that suppressing the norm of this Jacobian helps in bounding quantization-induced performance degradation. Based on this, we propose Jacobian-Guided Noise Injection, a training strategy that injects zero-mean Gaussian noise into pre-attention logits, with variance derived directly from the Jacobian Frobenius norm. Unlike prior approaches that rely on heuristic or penalise jacobian directly, our method provides a way to identify the optimal noise variance based on the local attention sensitivity. We evaluate the method on SOTA LLM architectures, where it demonstrates improved robustness over popular PTQ methods. Empirical analysis reveals that the proposed method gives up to +37% relative gains on Top-1 accuracy on ImageNet-1K for SigLIP and improves relative perplexity by upto 40% on WikiText for language models in low bit quantisation settings, proving the efficacy of the approach.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-47fec4c9cff83d45", "title": "TreeWY: Speculative Verification for Gated DeltaNet Hybrids", "authors": ["Sneha Murthy Ghantasala"], "year": 2026, "published": "2026-08-21T10:31:15Z", "updated": "2026-08-21T10:31:15Z", "arxiv_id": "http://arxiv.org/abs/2608.20961v1", "url": "https://arxiv.org/abs/2608.20961v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CL", "cs.DC", "cs.LG", "cs.PF"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache", "Speculative Decoding", "Memory / RAG"], "benchmarks": [], "models": ["Qwen3.5", "ying"], "key_results": ["Modern open models are hybrids: most layers are linear-attention (Gated DeltaNet, GDN) layers carrying a small fixed-size recurrent state instead of a growing key-value (KV) cache."], "abstract": "Modern open models are hybrids: most layers are linear-attention (Gated DeltaNet, GDN) layers carrying a small fixed-size recurrent state instead of a growing key-value (KV) cache. This makes ordinary decoding memory-efficient, but hurts speculative decoding. To verify a batch of draft tokens and then roll back the rejected ones, today's systems snapshot the full recurrent state at every draft position for GDN layers, and those snapshots cannot be shared across branches of a draft tree, so a wide, high-acceptance tree becomes memory-infeasible. We remove the snapshots. Using a tree-structured WY transform of the gated delta rule, we compute every draft node's output with a single triangular solve and reconstruct only the one accepted state on commit, storing a small pseudo-value matrix instead of per-node states; the derivation depends only on the gated delta rule, not on any other architectural detail. In serving benchmarks on two scales of one hybrid model family (Qwen3.5 35B and 397B) this cuts speculative recurrent-state memory and KV-cache pressure at identical acceptance length, turning the freed HBM into higher throughput and much lower time-to-first-token (TTFT) wherever memory binds, and costing a few percent where it does not. For tree width the same memory buys affordability: a wider, higher-acceptance draft becomes possible, though not yet a throughput win.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-a455a17ecd3ad903", "title": "Quantization-Aware Healing: A Practical Recipe for Recovering Compressed, 4-Bit LLMs", "authors": ["Bakbergen Ryskulov", "Iker García-Ferrero", "David Montero", "David Jansen", "Ali Hashemi", "Jezabel R. Garcia", "Antonio Tiene", "Román Orús"], "year": 2026, "published": "2026-08-21T10:19:27Z", "updated": "2026-08-21T10:19:27Z", "arxiv_id": "http://arxiv.org/abs/2608.20953v1", "url": "https://arxiv.org/abs/2608.20953v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.LG", "cs.PF"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Quantization", "Long Context", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["On a GPT-OSS 120B to 60B to MXFP4 pipeline, the QAH student matches or beats its bfloat16 source on 7 of 9 benchmarks at roughly 4 times less weight memory and half the teacher's parameter count, and is released open-weight as Hypernova-60B.", "Against a matched QAT baseline it reaches a comparable peak about 7 times faster and stays stable under continued training, without hand-tuned early stopping."], "abstract": "Serving large language models cheaply increasingly means shipping models that are both structurally compressed to a fraction of their parameters and quantized to 4 bits. Together these steps degrade reasoning, mathematics, coding, and long-context behavior enough to require a recovery, or healing, stage before deployment. The default recipe, quantization-aware training (QAT), re-fits the compressed, quantized model to hard labels; in our pipeline it converged slowly and collapsed past its peak. We adopted Quantization-Aware Healing (QAH) instead. Because a structurally compressed model is never independently trained at full precision, its bfloat16 checkpoint is a distillation-recovered approximation of the original; QAH distills the 4-bit student directly from the original, uncompressed model. On a GPT-OSS 120B to 60B to MXFP4 pipeline, the QAH student matches or beats its bfloat16 source on 7 of 9 benchmarks at roughly 4 times less weight memory and half the teacher's parameter count, and is released open-weight as Hypernova-60B. Against a matched QAT baseline it reaches a comparable peak about 7 times faster and stays stable under continued training, without hand-tuned early stopping. We also report deployment lessons, including a large, reproducible quality gap between distributed-training backends. Our aim is a recipe deployable without a multi-week hyper-parameter search.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-6359b9c917b57fbf", "title": "MentorPulse: Refreshing Cross-Model Latent Guidance for Long-Form Generation", "authors": ["Ziwu Liu", "Guozhong Li", "Chen Qiu", "Weiyang Kong", "Panos Kalnis"], "year": 2026, "published": "2026-08-21T09:49:09Z", "updated": "2026-08-21T09:49:09Z", "arxiv_id": "http://arxiv.org/abs/2608.20927v1", "url": "https://arxiv.org/abs/2608.20927v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["KV Cache", "Memory / RAG"], "benchmarks": [], "models": [], "key_results": ["Across thirteen datasets, MentorPulse closes 52.2% of the mentor-student gap on macro average, outperforming C2C, T2T, and equal-budget LoRA, with the largest gains on long outputs."], "abstract": "Cross-model latent guidance lets a frozen large mentor encode an input once and a frozen small student generate from the resulting signal. Existing methods keep this signal fixed, assuming it stays useful as the output grows; we show this fails in long-form generation. On multi-turn instruction following, static guidance pushes a 4B student's constraint satisfaction 2.5 points below its no-guidance baseline; a training-free refresh every 16 tokens changes only the memory content and restores a 2.0-point gain over that baseline. We propose MentorPulse to keep guidance fresh at practical cost: it compresses mentor states into a capped slot memory, incrementally processes newly generated tokens, and updates the memory that the student reads through gated cross-attention without resetting the student's KV cache. Windowed Refresh Training exposes the bridge to prefix-conditioned memory. Across thirteen datasets, MentorPulse closes 52.2% of the mentor-student gap on macro average, outperforming C2C, T2T, and equal-budget LoRA, with the largest gains on long outputs. It performs best on all eleven mentor-student pairs from three model families, with margins that narrow as the capability gap grows, and a lightweight read-pattern check predicts the gain before deployment. Measured costs identify refresh intervals that dominate text guidance on long outputs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-d7ff360e51a97fee", "title": "ForeDreamer: A Self-Evolving Dual-Agent Memory Architecture for Future Event Prediction", "authors": ["Linhao Zhong", "Zongze Du", "Linyu Wu", "Yu Bo", "Hourong Li", "Chenchen Jing", "Hao Chen", "Yuling Xi", "Chunhua Shen"], "year": 2026, "published": "2026-08-21T09:38:27Z", "updated": "2026-08-21T09:38:27Z", "arxiv_id": "http://arxiv.org/abs/2608.20920v1", "url": "https://arxiv.org/abs/2608.20920v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "Efficient Training & Inference", "methods": ["Distillation", "Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Open-web future event prediction requires agents to distill reliable signals from noisy, redundant, and incomplete evidence."], "abstract": "Open-web future event prediction requires agents to distill reliable signals from noisy, redundant, and incomplete evidence. Existing retrieval/memory mechanisms directly feed retrieved information to agents or rely on simple memory functions such as storing and reusing prior information for prediction, leaving them insufficient for open-web forecasting. We propose to transform raw web evidence into structured memory before prediction, enabling agents to reason over distilled, question-specific evidence rather than noisy retrieval results. This paper presents ForeDreamer, a self-evolving dual-agent framework for managing memory over open-web evidence. ForeDreamer separates factual memory, a question-specific evidence state for the current forecast, from experiential memory, persistent agent experience accumulated across forecasting episodes. It uses a main agent for search and prediction, and a memory-processing subagent to convert search results into factual memory with dedicated tools. ForeDreamer further evolves experiential memory through two tracks, improving both forecasting decisions and factual-memory construction. Experiments on Prophet Arena and FutureX demonstrate the effectiveness of ForeDreamer. Project page: https://zhongzero.github.io/ForeDreamer", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-4beac25bb5834027", "title": "EviRank: Structured Relevance Evidence for Multimodal Image Re-ranking", "authors": ["Enjun Du", "Siyi Liu", "Zirong Chen", "Xinyu Zuo", "Jinwen Luo", "Ruiwen Tao", "Lisheng Duan", "Haijin Liang", "Jin Ma", "Junfu Pu", "Yongqi Zhang"], "year": 2026, "published": "2026-08-21T09:04:39Z", "updated": "2026-08-21T09:04:39Z", "arxiv_id": "http://arxiv.org/abs/2608.20886v1", "url": "https://arxiv.org/abs/2608.20886v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Chain-of-Thought", "Distillation", "VLM"], "benchmarks": [], "models": [], "key_results": ["Re-ranking then reduces to evidence-conditioned verification, combining deterministic rubric scoring and evidence-grounded listwise comparison in a single training-free procedure.", "Across five benchmarks spanning text-to-image, image-to-image, and composed image retrieval, EviRank achieves state-of-the-art performance, and the distilled student preserves over 90% of the teacher's capability at substantially lower cost."], "abstract": "Real-world image search queries are multimodal and compositional: ``find this shirt in pink'' specifies an entity to retain, an attribute to modify, and context to ignore. Yet existing re-rankers either compress such multifaceted relevance into an opaque embedding or rely on free-form chain-of-thought that easily omits or hallucinates fine-grained constraints. Drawing on rubric- and checklist-based evaluation from NLP, we recast multimodal image re-ranking as a semantic constraint satisfaction problem and propose EviRank, which parses any query - text-only, image-only, or composed - into a unified evidence package: typed criteria across six semantic slots (e.g., entities, attributes, relations), each labelled required, forbidden, or ignorable. Re-ranking then reduces to evidence-conditioned verification, combining deterministic rubric scoring and evidence-grounded listwise comparison in a single training-free procedure. The explicit evidence can further serve as structured supervision for optionally distilling a lightweight student. Across five benchmarks spanning text-to-image, image-to-image, and composed image retrieval, EviRank achieves state-of-the-art performance, and the distilled student preserves over 90% of the teacher's capability at substantially lower cost.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-19233c3a3c0d4dea", "title": "Nothing Changed but the Model: CellFill -- Bounded In-Cell Learning for Bit-Identical, Revocable Updates to Quantized LLMs", "authors": ["Zifeng Liu", "Zhiyong Du", "Yaxin Lu", "Yiming Mao", "Zhenhe Wang", "Wenqi Shi", "Zhengkun Jing"], "year": 2026, "published": "2026-08-21T08:49:07Z", "updated": "2026-08-21T08:49:07Z", "arxiv_id": "http://arxiv.org/abs/2608.20873v1", "url": "https://arxiv.org/abs/2608.20873v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Efficient Training & Inference", "methods": ["Quantization"], "benchmarks": [], "models": [], "key_results": ["Exact invariance turns out to be nearly free: across three paired seeds the constrained dense path matches an unconstrained reference whose weights provably escape the artifact (58.9 vs 59.3 percent fact recall; paired difference -0.5 points, 95% CI [-5.0,+4.0]), and is better on held-out cross-domain perplexity.", "Against the natural null hypothesis -- serving the same update as an unmerged adapter -- projecting into the cells reduces cross-domain forgetting in every run that converged, and a diverged control shows the boundary: projection is a trust region, not a repair."], "abstract": "Every way of teaching a deployed language model something new -- full fine-tuning, adapter merging, model editing -- replaces the released checkpoint, and with it every evaluation and cache that referred to those exact bits. We instead learn inside the dequantization gap: with the integer codes and scales of a 4-bit release frozen, new knowledge is written only into the per-weight residual that lives strictly inside each quantization decision cell. Re-quantization then returns the released artifact bit-for-bit, a machine-checkable guarantee; updates are exactly revocable by dropping the residual; and drift is bounded. We give six propositions and three training paths, including CellFill, a bounded reparameterization that makes invariance structural rather than enforced.\n  Exact invariance turns out to be nearly free: across three paired seeds the constrained dense path matches an unconstrained reference whose weights provably escape the artifact (58.9 vs 59.3 percent fact recall; paired difference -0.5 points, 95% CI [-5.0,+4.0]), and is better on held-out cross-domain perplexity. Against the natural null hypothesis -- serving the same update as an unmerged adapter -- projecting into the cells reduces cross-domain forgetting in every run that converged, and a diverged control shows the boundary: projection is a trust region, not a repair. What no method escapes is the cost of knowledge itself, and the apparent free lunch of in-domain perplexity improving past the anchor is an artifact of rehearsal sharing a corpus with the metric. Methods differ threefold at matched rehearsal in knowledge bought per point of cross-domain perplexity, a ranking that is not the recall ranking. The method transfers to a 27B hybrid linear-attention model (2.4e10 constrained weights, verified bit-identical), where matched recall costs about half as much cross-domain perplexity as at 1.7B.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-57295a34f9c6ffa5", "title": "STAR-OPD: Structured Aspect-Cascade-Aware On-Policy Reward Distillation for ABSA Quadruple Extraction", "authors": ["Tong Sun", "Mingyang Ma", "Jiayang Yu"], "year": 2026, "published": "2026-08-21T07:50:00Z", "updated": "2026-08-21T07:50:00Z", "arxiv_id": "http://arxiv.org/abs/2608.20831v1", "url": "https://arxiv.org/abs/2608.20831v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "Efficient Training & Inference", "methods": ["Chain-of-Thought", "Distillation"], "benchmarks": [], "models": ["Qwen3-4B"], "key_results": ["Experiments on E-ABSA20K and SemEval-2014 show that STAR-OPD consistently outperforms off-policy and general on-policy baselines, reduces target hallucination, and substantially improves performance on structurally hard cases."], "abstract": "Aspect-based sentiment analysis (ABSA) quadruple extraction requires jointly predicting target, aspect, opinion, and sentiment over reviews that often contain multiple fine-grained sentiment tuples. While large chain-of-thought (CoT) models perform well on this task, distilling them into smaller deployable models remains difficult. We identify a task-specific failure mode in distilled ABSA extraction: student errors at the target-aspect interface create structurally invalid states, such as broken target-aspect bindings and hallucinated targets, which then corrupt downstream predictions. Conventional off-policy distillation is poorly suited to this setting because it trains only on teacher-generated trajectories and provides little supervision on the student-induced structural states that dominate inference. To address this mismatch, we propose STAR-OPD (STructured Aspect-cascade-aware On-Policy Reward Distillation), which builds on generic on-policy distillation and instantiates it for ABSA quadruple extraction with cascade-aware, set-structured rewards. STAR-OPD trains on student rollouts and applies set-structured rewards that directly target binding consistency, target grounding, and fine-grained aspect disambiguation. Experiments on E-ABSA20K and SemEval-2014 show that STAR-OPD consistently outperforms off-policy and general on-policy baselines, reduces target hallucination, and substantially improves performance on structurally hard cases. With Qwen3-4B, STAR-OPD substantially narrows the student-teacher gap while improving inference efficiency, highlighting the importance of on-policy structural correction for distilled ABSA extraction.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-3fe64e536dd65a2b", "title": "Scaling Muon for Diffusion Transformers", "authors": ["Chenghao Li", "Xiao Han", "Xinxin Huang", "Wei Liu", "Boyang Li", "Bing Xiao", "Heran Zhang", "Juanma Perez Rua", "Ke Xu", "Kangning Liu", "Linjun Kuang", "Na Li", "Tan Wang", "Tian Xie", "Wei Peng", "Yang Pei", "Yifan Xu", "Yuanhao Zhai", "Yuwei Lin", "Zhe Wang", "Zihao He", "Daniel Li", "Junbiao Tang", "Ziyang Jiang", "Dake Chen"], "year": 2026, "published": "2026-08-21T07:40:28Z", "updated": "2026-08-21T07:40:28Z", "arxiv_id": "http://arxiv.org/abs/2608.20818v1", "url": "https://arxiv.org/abs/2608.20818v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI", "cs.CV"], "pillar": "Efficient Training & Inference", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["The matrix-aware optimizer Muon improves large model training by balancing updates across singular directions, yet its scaling behavior and end-to-end efficiency on large Diffusion Transformers (DiTs) remain unclear.", "Across all scales, Muon improves the best observed generative quality over AdamW by 12.9--19.1\\%.", "Compared with vanilla Muon, Periodic Row-wise Muon remains within 0.5\\% in best generative quality on the 1.3B--4B models and improves it by 4.5\\% at 9B."], "abstract": "The matrix-aware optimizer Muon improves large model training by balancing updates across singular directions, yet its scaling behavior and end-to-end efficiency on large Diffusion Transformers (DiTs) remain unclear. We first establish Muon's scaling behavior on DiTs from 1.3B to 15B parameters, showing that its optimization and generative quality advantages over AdamW persist across model scales. However, at scale, the 5-step Newton--Schulz iteration (NS5) performed at every optimization step, together with full-momentum materialization, introduces substantial computation and communication overhead that can offset Muon's step-efficiency advantage. We introduce \\emph{Periodic Row-wise Muon}, which performs a full NS5 spectral update once every \\(K\\) steps and applies a low compute and communication cost row-wise constrained update based on the current momentum at the remaining steps. We further co-design a distributed implementation that operates directly on sharded momentum during non-refresh steps and accelerates spectral refreshes through bucketed all-gather and communication--computation overlap. Across all scales, Muon improves the best observed generative quality over AdamW by 12.9--19.1\\%. Compared with vanilla Muon, Periodic Row-wise Muon remains within 0.5\\% in best generative quality on the 1.3B--4B models and improves it by 4.5\\% at 9B. It reduces optimizer time by 46.9--54.3\\%, end-to-end step time by 15.7--24.3\\%, and logical communication volume by 66.7\\%, while reaching its respective best generative quality with 33.7--64.8\\% less active training time. These results show that Periodic Row-wise Muon preserves Muon's generative quality advantage while translating it into end-to-end training efficiency for large DiTs.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-cfe8c15c22f6d65d", "title": "Truthful Calibration Measures for Sequential Prediction", "authors": ["Anagha Gokul", "Jason Hartline", "Lunjia Hu", "Jonathan Ullman", "Yifan Wu"], "year": 2026, "published": "2026-08-21T17:54:05Z", "updated": "2026-08-21T17:54:05Z", "arxiv_id": "http://arxiv.org/abs/2608.21348v1", "url": "https://arxiv.org/abs/2608.21348v1", "primary_category": "cs.DS", "categories": ["cs.DS", "cs.GT", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability"], "benchmarks": [], "models": ["ying"], "key_results": ["This improves the approximate-truthfulness guarantee of Haghtalab et al."], "abstract": "Calibration requires probabilistic reports to be conditionally unbiased and reliably interpretable as probabilities. A calibration measure assigns numerical error to miscalibrated reports. Haghtalab et al. (2024) proposed an approximately truthful calibration measure for online prediction, leaving open whether exact truthfulness is compatible with completeness and soundness.\n  We resolve this question negatively for sequential binary prediction: exact truthfulness is incompatible with completeness and soundness, even for independent outcomes. We then show that this impossibility is specific to exact truthfulness. We give two general reductions from a base calibration measure, producing additively and multiplicatively approximately truthful calibration measures, respectively. Applying the multiplicative reduction, for every $0 < \\varepsilon < 1$ we construct a sound and complete calibration measure that is $(1+\\exp(-T^{(1-\\varepsilon)/2}/2))$-multiplicatively truthful. This improves the approximate-truthfulness guarantee of Haghtalab et al. (2024).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-56a9d7145ba46fe4", "title": "Move by Move: Measuring and Steering How LLMs Conduct Psychotherapy", "authors": ["Afonso Baldo", "Hugo Pitorro", "Areti Vassilopoulos", "Anabela C. Areias", "Maya D'Eon", "Fabíola Costa", "Ricardo Rei", "Nuno M. Guerreiro"], "year": 2026, "published": "2026-08-21T17:32:38Z", "updated": "2026-08-21T17:32:38Z", "arxiv_id": "http://arxiv.org/abs/2608.21325v1", "url": "https://arxiv.org/abs/2608.21325v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["We introduce an ontology of ten therapeutic moves: compact, function-based categories grounded in the MULTI-60 inventory, validated through an annotation campaign with five licensed psychologists, and scaled with a judge-based approach that matches expert agreement.", "Exposing the ontology as a set of tools roughly halves the mean deviation from the human move distribution and improves turn-level alignment with human therapist by 7-9 percentage points, without any fine-tuning."], "abstract": "Users increasingly turn to large language models for emotional support, yet little is known about how these models actually conduct a psychotherapy interaction. We introduce an ontology of ten therapeutic moves: compact, function-based categories grounded in the MULTI-60 inventory, validated through an annotation campaign with five licensed psychologists, and scaled with a judge-based approach that matches expert agreement. Applying it to real counseling transcripts and model-led sessions, we compare the move distributions between human clinicians and a panel of frontier models. Models over-use inquiry at up to three times the human rate, neglect psychoeducation, and are strongly context-anchored: they carry forward strategies initiated by a human clinician but rarely initiate them themselves. Exposing the ontology as a set of tools roughly halves the mean deviation from the human move distribution and improves turn-level alignment with human therapist by 7-9 percentage points, without any fine-tuning.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-b990b68f9ac55787", "title": "ConceptTS: LLM-Guided Concept Bottlenecks for Interpretable Multivariate Time-Series Forecasting", "authors": ["Yichen Jiang", "Yueqiao Chen", "Dongyu Liu"], "year": 2026, "published": "2026-08-21T16:35:44Z", "updated": "2026-08-21T16:35:44Z", "arxiv_id": "http://arxiv.org/abs/2608.21277v1", "url": "https://arxiv.org/abs/2608.21277v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Interpretability", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["State-of-the-art multivariate time-series forecasters can model complex temporal and cross-variable dependencies, yet their opaque representations provide limited insight into why a particular forecast is produced.", "Experiments on the Beijing Multi-Site Air Quality dataset show that ConceptTS achieves accuracy competitive with strong black-box baselines while producing semantically meaningful concept activations."], "abstract": "State-of-the-art multivariate time-series forecasters can model complex temporal and cross-variable dependencies, yet their opaque representations provide limited insight into why a particular forecast is produced. This lack of transparency restricts their use in settings where practitioners must understand and assess the factors underlying a prediction. We introduce ConceptTS, an interpretable forecasting framework that organizes its predictions around named, human-readable concepts. ConceptTS uses a large language model to propose task-relevant concepts and generate executable labeling rules, translating the language model's domain knowledge into direct supervision without costly manual concept annotation. The proposed concepts are organized into three complementary bottlenecks that describe the historical context, local forecast intervals, and the full forecast horizon. A shared decoder combines representations derived from their predicted activations to construct the forecast, making the model's decision process explicit and supporting direct concept-level interventions. Experiments on the Beijing Multi-Site Air Quality dataset show that ConceptTS achieves accuracy competitive with strong black-box baselines while producing semantically meaningful concept activations.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-465370d94a9618f1", "title": "TRACE-C: Rank-Calibrated Relational Anomaly Detection for Multi-Stream Operational Telemetry", "authors": ["Matthew Faucher"], "year": 2026, "published": "2026-08-21T16:01:56Z", "updated": "2026-08-21T16:01:56Z", "arxiv_id": "http://arxiv.org/abs/2608.21251v1", "url": "https://arxiv.org/abs/2608.21251v1", "primary_category": "cs.LG", "categories": ["cs.LG", "stat.ML"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["TRACE-C is an auditable strictly-prior rank-calibrated detector for aligned multi-stream telemetry: same-regime rolling median/MAD residuals feed three window channels -- a maximum normalized local sum, a Gaussian copula-form dependence contrast on robust-z residuals, and a worst standardized AR(1) innovation -- whose channel ranks are Fisher-aggregated and ranked against earlier aggregates."], "abstract": "Operational telemetry can be jointly anomalous while every individual stream stays inside its familiar range. TRACE-C is an auditable strictly-prior rank-calibrated detector for aligned multi-stream telemetry: same-regime rolling median/MAD residuals feed three window channels -- a maximum normalized local sum, a Gaussian copula-form dependence contrast on robust-z residuals, and a worst standardized AR(1) innovation -- whose channel ranks are Fisher-aggregated and ranked against earlier aggregates.\n  We evaluate six Great Britain grid streams with a January-April 2019 fit, July-December 2019 development evidence, and a 2020 hold-out frozen before inspection. TRACE-C ranks Storm Atiyah first among 2019 test windows, but a disclosed channel ablation attributes that rank to the local channel, not the copula-form channel: copula-only ranks Atiyah 59th. The short 9 August frequency event is ranked far lower by the fused detector (143) than by the temporal channel alone (40), and reconstruction baselines rank it first. In 2020 no window is selected, which is consistent with record-rule saturation rather than an uneventful year; the highest-ranked frozen window was later interpreted as Storm Ellen.\n  Three interpretive limits carry throughout. The resulting p-values are selection quantities, not event probabilities. The copula-form channel is not a literal copula density: the method applies no probability-integral or normal-score transform. Empirical rank counts are diagnostics, not coverage or false-discovery proofs. Every table and figure in this paper is generated from committed machine-readable reports.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-765ac3598eba5596", "title": "COEC: Calibrated Orthogonal-Equivalence Compensation for Structured Pruning of Large Language Models", "authors": ["Peiqi Yu", "Nam Ling", "Wei Wang", "Wei Jiang"], "year": 2026, "published": "2026-08-21T14:20:49Z", "updated": "2026-08-21T14:20:49Z", "arxiv_id": "http://arxiv.org/abs/2608.21142v1", "url": "https://arxiv.org/abs/2608.21142v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["Llama-3", "Llama-3.1", "Qwen2.5"], "key_results": ["Structured pruning reduces the size and inference cost of large language models (LLMs) by removing weight columns, but the resulting output error can degrade accuracy.", "Experiments on the Llama-3, Llama-3.1, and Qwen2.5 model families across multiple structured sparsity levels show that COEC improves perplexity on every model and zero-shot accuracy in most settings over existing compensation methods, with larger gains at higher sparsity."], "abstract": "Structured pruning reduces the size and inference cost of large language models (LLMs) by removing weight columns, but the resulting output error can degrade accuracy. Existing training-free compensation methods use an additive bias or a single orthogonal rotation on the output side of the retained weight. These corrections leave its input singular frame unchanged and therefore limit how the retained weight can adapt after column removal. We propose COEC (Calibrated Orthogonal-Equivalence Compensation), a training-free compensation framework that applies alternating left and right orthogonal rotations to the retained weight. The right rotation is optimized on a reduced Stiefel manifold, while singular values are rescaled using generalized cross-validation to select the regularization strength for each layer. COEC further tempers the calibration Gram matrix to reduce the dominance of high-energy activation directions and introduces an alignment penalty that preserves the geometric relation between adjacent attention projections.All components use second-order statistics from a small calibration set and require neither backpropagation through the LLM nor retraining of the model parameters. COEC is independent of the column pruning criterion and can be applied to multiple structured pruning methods. Experiments on the Llama-3, Llama-3.1, and Qwen2.5 model families across multiple structured sparsity levels show that COEC improves perplexity on every model and zero-shot accuracy in most settings over existing compensation methods, with larger gains at higher sparsity. These results show that post-pruning compensation can recover part of the performance lost to column removal.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-23dda143562d42f6", "title": "PromptResponse: Optimizing Prompts for LLM Coding Tasks", "authors": ["Erik Thureck", "Robert Kühnen", "Tim Jacobowitz"], "year": 2026, "published": "2026-08-21T13:16:48Z", "updated": "2026-08-21T13:16:48Z", "arxiv_id": "http://arxiv.org/abs/2608.21074v1", "url": "https://arxiv.org/abs/2608.21074v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI", "cs.HC", "cs.SE"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": ["HumanEval"], "models": ["GPT-4o", "yield"], "key_results": ["Our results show that consistent formatting$\\unicode{x2014}$especially JSON$\\unicode{x2014}$improves generation efficiency and syntactic stability, with minor gains in task performance.", "Conversely, the LLM-tuned prompts resulted in significantly degraded task performance without significant improvements in any other dimension.", "These findings suggest that low-effort reformatting alone can yield measurable improvements, while tuning must account for model alignment."], "abstract": "Large language models (LLMs) are increasingly used in research workflows and software development pipelines, yet their output remains sensitive to input prompt variations. This paper presents $\\unicode{x00AB}$PromptResponse$\\unicode{x00BB}$, a controlled study examining how formatting and LLM-based tuning of coding task prompts affect the resulting code's performance, efficiency, and stability. Using five semantically identical yet syntactically distinct variants of the HumanEval dataset$\\unicode{x2014}$baseline, JSON, Markdown, YAML, and an LLM-tuned version$\\unicode{x2014}$we had GPT-4o solve its coding problems over 8200$\\unicode{x00A0}$executions. Our results show that consistent formatting$\\unicode{x2014}$especially JSON$\\unicode{x2014}$improves generation efficiency and syntactic stability, with minor gains in task performance. Conversely, the LLM-tuned prompts resulted in significantly degraded task performance without significant improvements in any other dimension. These findings suggest that low-effort reformatting alone can yield measurable improvements, while tuning must account for model alignment. We conclude our work with providing a set of practical recommendations informed by our results as well as releasing our dataset variants and evaluation pipeline for future work.", "code_available": false, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-bebf8a6633a10e77", "title": "Designing a Robust LLM-Based Evaluation System for Agentic AI in Drug Discovery Through Human Alignment", "authors": ["Emma Granqvist", "Rocío Mercado", "Samuel Genheden"], "year": 2026, "published": "2026-08-21T12:58:24Z", "updated": "2026-08-21T12:58:24Z", "arxiv_id": "http://arxiv.org/abs/2608.21057v1", "url": "https://arxiv.org/abs/2608.21057v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization", "Multi-Agent"], "benchmarks": [], "models": ["GPT-5", "Claude Opus 4.7", "Gemini 3.1 Pro", "Llama 3.1", "ying"], "key_results": ["Agentic large language model (LLM) systems are reshaping scientific workflows in chemistry and drug discovery, but evaluating their open-ended, tool-augmented outputs remains a fundamental bottleneck."], "abstract": "Agentic large language model (LLM) systems are reshaping scientific workflows in chemistry and drug discovery, but evaluating their open-ended, tool-augmented outputs remains a fundamental bottleneck. Reference-based metrics such as BLEU and ROUGE fail to capture semantic correctness, while expert human evaluation does not scale to the iteration speed these systems demand. The LLM-as-a-Judge paradigm has emerged as a scalable alternative, but existing drug discovery benchmarks deploy LLM judges without validating their alignment with human experts. In this work, we present an LLM-as-a-Judge evaluation framework for ChatInvent, an agentic drug discovery assistant deployed at AstraZeneca, with four contributions. First, we define four output-quality evaluation dimensions---Completeness, Relevancy, Structural Clarity, and Scope Adherence---alongside deterministic Tool Call Correctness checks. Second, we validate the judge through a human alignment study with five expert annotators, comparing Gemini 3.1 Pro, Claude Opus 4.7, GPT-5, and Llama 3.1 70B as candidate judges. Third, we optimize the best-performing judge using few-shot demonstrations of human-annotated examples, improving alignment with the human majority vote from 0.80 to 0.86. Fourth, applying the optimized judge to 70 held-out questions, we surface concrete limitations and find that informal phrasings do not systematically degrade output quality; if anything, it is helpful to have the LLM rewrite the original question before querying the agent. Our framework provides a reusable template for human-aligned evaluation of agentic systems in scientific domains.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-ad642d00faa8e184", "title": "Evidence-Consistent Generative Detection under Scenario-Level Distribution Shift", "authors": ["San Kim", "JinYeong Bak"], "year": 2026, "published": "2026-08-21T12:36:36Z", "updated": "2026-08-21T12:36:36Z", "arxiv_id": "http://arxiv.org/abs/2608.21043v1", "url": "https://arxiv.org/abs/2608.21043v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": ["phishing"], "key_results": ["On the 0.5B decoder, relative to the same backbone trained without consistency regularization, ECoG raises Macro-F1 on OOD challenging instances by 3.22 points, reduces the share of predictions whose generated rationale supports the opposite label by 4.22 points, and increases token-level overlap with reference evidence spans by 8.38 points; the reduction in prediction-rationale inconsistency is consistent across four decoder backbones."], "abstract": "Conventional in-distribution evaluation can overestimate robustness when training and test data share recurring task-specific patterns or surface cues. This risk is especially relevant in social-engineering fraud detection, where attackers can preserve malicious intent while changing the scenario, impersonated entity, or wording. We study this problem as scenario-level out-of-distribution (SL-OOD) detection for SMS and voice phishing, where entire attack scenarios are held out from training while the label space remains fixed. This setting tests whether models can generalize to unseen attack scenarios using decision-relevant evidence rather than familiar scenario-specific cues. Using this SL-OOD evaluation, we find that high in-distribution performance does not reliably predict held-out robustness across feature-, encoder-, and decoder-based baselines. We interpret this gap as scenario memorization: reliance on recurring scenario-specific lexical or entity cues rather than decision-relevant evidence. We propose ECoG, an evidence-consistent generative framework that combines evidence-span supervision with a rationale-label consistency objective during training. On the 0.5B decoder, relative to the same backbone trained without consistency regularization, ECoG raises Macro-F1 on OOD challenging instances by 3.22 points, reduces the share of predictions whose generated rationale supports the opposite label by 4.22 points, and increases token-level overlap with reference evidence spans by 8.38 points; the reduction in prediction-rationale inconsistency is consistent across four decoder backbones. These results suggest that compact generative detectors can benefit from evidence supervision and rationale-label consistency under social-engineering shift.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-45a25bd614c246cc", "title": "Scaling Unsupervised Word Alignment to Documents via Structural Constraints", "authors": ["Michelle Wastl", "Jannis Vamvas", "Rico Sennrich"], "year": 2026, "published": "2026-08-21T12:13:21Z", "updated": "2026-08-21T12:13:21Z", "arxiv_id": "http://arxiv.org/abs/2608.21023v1", "url": "https://arxiv.org/abs/2608.21023v1", "primary_category": "cs.CL", "categories": ["cs.CL"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["ying"], "key_results": ["Averaged over three models, CTFAlign reduces word alignment error rate from 0.412 to 0.326.", "These gains transfer downstream, leading to improvements in document-level translation coverage evaluation and recognition of semantic differences."], "abstract": "Word alignment has traditionally been studied between sentences, but many cross-lingual tasks increasingly require correspondences across full documents. While recent multilingual embedding models can encode long inputs, we show that applying algorithms designed for sentences directly to documents leads to performance degradation. To address this, we introduce CTFAlign, a lightweight, training-free approach for document-level word alignment. CTFAlign applies a coarse-to-fine refinement strategy that restricts the alignment search space to semantically similar regions. Additionally, we introduce MDPAlign, a simpler alternative that constrains alignments by position with a main diagonal prior. Both approaches operate directly on full documents without relying on sentence segmentation or sentence alignment. We evaluate these methods across six language pairs varying in typological distance, resourcedness, and document length. Averaged over three models, CTFAlign reduces word alignment error rate from 0.412 to 0.326. These gains transfer downstream, leading to improvements in document-level translation coverage evaluation and recognition of semantic differences. We release CTFAlign as a Python package and make the code and data to reproduce our experiments publicly available.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-7ea44888c7e290af", "title": "Trojaning the Alignment: Stealthy Backdoor Attacks against Graph Foundation Models", "authors": ["Minhua Lin", "Zhicheng Gao", "Yilong Wang", "Hanqing Lu", "Xiang Zhang", "Suhang Wang"], "year": 2026, "published": "2026-08-21T11:20:35Z", "updated": "2026-08-21T11:20:35Z", "arxiv_id": "http://arxiv.org/abs/2608.20991v1", "url": "https://arxiv.org/abs/2608.20991v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["r1c"], "key_results": ["Graph Foundation Models (GFMs) on text-attributed graphs (TAGs) align graph representations with language semantics to support transferable graph learning."], "abstract": "Graph Foundation Models (GFMs) on text-attributed graphs (TAGs) align graph representations with language semantics to support transferable graph learning. Despite these advantages, the backdoor vulnerability of GFMs on TAGs remains insufficiently understood, especially under graph-language alignment, where graph and text representations are trained to constrain each other in a shared semantic space. Existing backdoor attacks mainly target either the graph side or the text side, treating the two modalities independently. This makes direct adaptation ineffective: graph-only triggers can be constrained by clean text semantics, while text-only triggers alter the language view but do not directly shift the graph representation being aligned and scored. TAGs also impose a stealth challenge because triggers are exposed as both node text and local graph structure, making incoherent trigger attributes or anomalous subgraphs easy to inspect or filter. In this paper, we propose STAG, a stealthy trojan attack framework designed for the graph-language alignment interface of GFMs on TAGs. STAG coordinates a graph-trigger generator with a text-side soft prompt so that trigger-attached graph representations and triggered text representations move toward the same target-class text region. To address TAG-specific stealthiness, STAG realizes trigger nodes as readable text through candidate retrieval and regularizes the trigger-attached subgraph so that its local structure remains close to the original subgraph. Extensive experiments on multiple TAG datasets and representative GFMs demonstrate the effectiveness and stealthiness of STAG. Our code is available at https://github.com/ventr1c/STAG.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-fe8da9a09c2ab495", "title": "KREL: Automatic Medical Coding via Knowledge-Guided Reasoning over Clinical Evidence with LLMs", "authors": ["Xubin Chen", "Yipeng Zhou", "Wen Sun", "Chengkai Huang", "Xiaoming Fu", "Quan Z. Sheng"], "year": 2026, "published": "2026-08-21T09:05:15Z", "updated": "2026-08-21T09:05:15Z", "arxiv_id": "http://arxiv.org/abs/2608.20887v1", "url": "https://arxiv.org/abs/2608.20887v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.AI"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["Experiments on benchmark datasets show that KREL consistently outperforms strong PLM-based and state-of-the-art LLM-based baselines."], "abstract": "Automatic Medical Coding (AMC), which assigns standardized International Classification of Diseases (ICD) codes to clinical notes, is essential for medical reimbursement, quality reporting, and clinical research. Existing pre-trained language model (PLM)-based methods typically formulate AMC as an extreme multi-label classification problem over a predefined code set, while recent large language model (LLM)-based approaches instead frame it as generation or multi-step reasoning. However, key challenges remain, including the extreme length of clinical notes that hinders effective interpretation, the vast ICD label space, and complex coding rules that are not explicitly captured by LLMs. In this work, we propose Knowledge-Guided Reasoning over Clinical Evidence with LLMs (KREL), a framework that leverages LLMs for clinical text understanding and reasoning while integrating external ICD coding guidelines as structured knowledge. This design enables tight coupling between domain knowledge and LLM reasoning, reducing hallucinations and improving compliance with coding standards. Experiments on benchmark datasets show that KREL consistently outperforms strong PLM-based and state-of-the-art LLM-based baselines.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-cf5295b3a381f775", "title": "ReCurveflow: A Flow Matching Framework that Learns Curved Reaction Trajectories to Predict Transition State Geometries", "authors": ["Seungheun Baek", "Mogan Gim", "Jaewoo Kang"], "year": 2026, "published": "2026-08-21T08:38:30Z", "updated": "2026-08-21T08:38:30Z", "arxiv_id": "http://arxiv.org/abs/2608.20869v1", "url": "https://arxiv.org/abs/2608.20869v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": [], "benchmarks": [], "models": [], "key_results": ["We also introduce off-path correction, which grants ReCurveflow with the ability to produce corrective velocity fields when engaged off-path geometry states during inference rollout, leading to better resistance against exposure bias and accuracy in TS prediction.", "Across three data splits and six evaluation metrics, ReCurveflow achieves the best result on the majority of split-metric combinations against seven baselines."], "abstract": "Predicting transition states (TS) in chemical reactions is crucial, as they provide insights into reaction mechanisms. Recent work on TS prediction have focused on flow matching supervised on straight linear paths that do not align with actual reaction trajectories. We propose a novel flow matching-based framework ReCurveflow that learns to predict TS geometries supervised on continuously curved reference paths interpolated from a full NEB-derived band of molecular geometries. We also introduce off-path correction, which grants ReCurveflow with the ability to produce corrective velocity fields when engaged off-path geometry states during inference rollout, leading to better resistance against exposure bias and accuracy in TS prediction. Across three data splits and six evaluation metrics, ReCurveflow achieves the best result on the majority of split-metric combinations against seven baselines. Qualitative analyses further show that ReCurveflow generates reaction trajectories with energy profiles that closely track the reference NEB path, provides initializations that ease the NEB optimization bottleneck, and exhibits the intended corrective behavior in its learned velocity fields. The ReCurveflow codebase is publicly available at https://github.com/dmis-lab/ReCurveflow.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-3fc8c81998c66acc", "title": "SAC-Copula: Quality-Preserving Watermarking for Diffusion Language Models via Smooth Correlated Gumbel Fields", "authors": ["Baixin Li", "Haiyun He"], "year": 2026, "published": "2026-08-21T08:03:18Z", "updated": "2026-08-21T08:03:18Z", "arxiv_id": "http://arxiv.org/abs/2608.20839v1", "url": "https://arxiv.org/abs/2608.20839v1", "primary_category": "cs.CL", "categories": ["cs.CL", "cs.CR", "cs.LG"], "pillar": "RL / Alignment / Safety", "methods": ["Video Generation"], "benchmarks": [], "models": [], "key_results": ["Mechanism-level analysis shows that local correlation reduces latent perturbation roughness and better matches iterative refinement dynamics.", "Experiments on LLaDA show that SAC-Copula achieves a favorable quality-detectability trade-off compared with existing baselines.", "In particular, further evaluations on Dream-7B and additional datasets show that SAC-Copula substantially improves PPL tail stability over the i.i.d."], "abstract": "Watermarking diffusion language models (DLMs) requires mechanisms compatible with iterative parallel unmasking rather than autoregressive decoding. Existing sampling-based watermarking methods typically inject position-wise i.i.d. perturbations, which can be poorly aligned with DLM decoding dynamics and degrade generation quality. We propose SAC-Copula, a quality-preserving watermarking method for DLMs based on smooth, locally correlated Gumbel perturbation fields constructed via a Gaussian copula. We further develop a SAC-aware detector using covariance-aware filtering and native-sample calibration. Mechanism-level analysis shows that local correlation reduces latent perturbation roughness and better matches iterative refinement dynamics. Experiments on LLaDA show that SAC-Copula achieves a favorable quality-detectability trade-off compared with existing baselines. In particular, further evaluations on Dream-7B and additional datasets show that SAC-Copula substantially improves PPL tail stability over the i.i.d. Gumbel baseline, while maintaining strong low-FPR detectability and competitive overall generation quality. Additional token-edit stress tests further assess watermark robustness under controlled synchronization drift.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-4d4defab6198b776", "title": "VT-MUSE: Multimodal Unified Sequential Visuotactile Representation Learning for Manipulation", "authors": ["Congsheng Xu", "Qiaochu Yang", "Fangyuan Shi", "Yifan Han", "Baijun Chen", "Yiming Wang", "Haonan Zhao", "Daolin Ma", "Xiaokang Yang", "Hesheng Wang"], "year": 2026, "published": "2026-08-21T16:50:02Z", "updated": "2026-08-21T16:50:02Z", "arxiv_id": "http://arxiv.org/abs/2608.21290v1", "url": "https://arxiv.org/abs/2608.21290v1", "primary_category": "cs.RO", "categories": ["cs.RO", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["On the simulation benchmark, VT-MUSE outperforms the strongest baseline evaluated on all tasks by 11 percentage points and also achieves substantial improvements in real-world experiments."], "abstract": "We propose VT-MUSE, a Multimodal Unified SEquential representation learning framework for visuotactilemanipulation. Existing approaches often encode visual and tactile observations independently before fusion, limiting their ability to capture fine-grained cross-modal dependencies. Moreover, most methods focus on observations at the current time step and overlook the temporal evolution of contact. VT-MUSE addresses both limitations through a two-stage representation learning framework. In Stage I, modality specific encoders are jointly adapted via cross-modal temporal alignment and masked-view consistency. In Stage II, a conditional variational latent model processes masked visual sequences together with full tactile histories. Auxiliary decoders reconstruct the masked recent visual observations and predict tactile depth changes, encouraging the latent representation to retain both global visual context and local contact dynamics. The learned representation is subsequently integrated into a lightweight Transformer policy through gated cross-attention. On the simulation benchmark, VT-MUSE outperforms the strongest baseline evaluated on all tasks by 11 percentage points and also achieves substantial improvements in real-world experiments.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-0871676735932b2e", "title": "Just Noticeable Difference Modeling for Token Compression in Vision-Language-Action Models", "authors": ["Zhuoyuan Li", "Rui Zhao", "Jin Wang", "Hanwei Zhu", "Cong Zhang", "Giuseppe Valenzise", "Weisi Lin", "Kin-Man Lam"], "year": 2026, "published": "2026-08-21T15:59:37Z", "updated": "2026-08-21T15:59:37Z", "arxiv_id": "http://arxiv.org/abs/2608.21247v1", "url": "https://arxiv.org/abs/2608.21247v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.RO"], "pillar": "Multimodal / World Models", "methods": ["KV Cache", "VLM"], "benchmarks": [], "models": [], "key_results": ["Experiments on the LIBERO benchmark with OpenVLA and OpenVLA-OFT demonstrate that Action-JND consistently improves compression reliability, especially under aggressive compression ratios."], "abstract": "Token compression has become a key technique for reducing the inference cost of large foundation models, with approaches such as token pruning and KV-cache reuse widely adopted in vision-language models and recently explored for embodied agents. In embodied agents, tokens not only support perception and semantic understanding but also directly affect latency-sensitive closed-loop robot action prediction. Existing schemes typically guide compression using redundancy or importance cues, such as visual similarity, attention scores, and saliency. However, these cues only indirectly measure the key factor for safe compression: how much a token can change before causing an unacceptable deviation in downstream actions. This receiver-dependent tolerance is closely related to the principle of just noticeable difference (JND). Classical JND characterizes signal tolerance in the human visual system, while machine-oriented JND extends this concept to downstream machine responses. Building on this progression, we introduce Action-JND, which extends JND modeling to embodied perception by defining noticeability through the language-conditioned action response of a vision-language-action (VLA) policy in closed-loop control. A token change is considered admissible only when the induced action deviation remains within a tolerated margin. To realize this concept, we develop a lightweight token-wise JND estimator in deep visual-feature space to predict the maximum tolerable perturbation while preserving policy responses. The resulting action-tolerance score serves as a plug-and-play criterion for VLA compression paradigms, including stale-KV reuse and token pruning, prioritizing action-tolerant tokens for compression. Experiments on the LIBERO benchmark with OpenVLA and OpenVLA-OFT demonstrate that Action-JND consistently improves compression reliability, especially under aggressive compression ratios.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-0352ea099b4f2f08", "title": "A VLM Answer Is Not an Anomaly Score: Rank Compression in Training-Free Video Anomaly Detection", "authors": ["Inpyo Song", "Jangwon Lee"], "year": 2026, "published": "2026-08-21T15:56:00Z", "updated": "2026-08-21T15:56:00Z", "arxiv_id": "http://arxiv.org/abs/2608.21244v1", "url": "https://arxiv.org/abs/2608.21244v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Across four 7-8B VLMs, the probability readout outperforms the generated readout for every tested combination of answer scale, benchmark, and metric, with average gains ranging from 5 to 13 points across the four benchmark-metric pairs."], "abstract": "Vision-language models enable training-free video anomaly detection by answering questions about video segments. VAD benchmarks, however, require a scalar anomaly score for each segment and evaluate the resulting ranking using the AUROC or AP. A VLM-based detector should therefore define an answer interface: the answer scale specifies the admissible answers, and the readout rule maps the model's output distribution to a score. Because this interface can change the evaluated ranking, it is part of the detector rather than a formatting detail. The generated readout uses only the most likely answer, whereas the probability readout uses the full distribution over admissible answers. Across four 7-8B VLMs, the probability readout outperforms the generated readout for every tested combination of answer scale, benchmark, and metric, with average gains ranging from 5 to 13 points across the four benchmark-metric pairs. The gap arises because the generated readout keeps only one answer value per segment, so segment with different answer distributions can receive the same score and lose their relative order. We call this loss of relative order generated-answer rank compression. Even when the answer scale allows 91 answers, the generated readout produces only 4-18 distinct scores, whereas the probability readout retains substantially finer score resolution. The advantage persists under every decoding strategy, prompt wording, and joint scoring-explanation prompt we test. The answer interface is therefore a consequential component of VLM-based VAD and should be explicitly specified and evaluated.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-318eac36388fe7d1", "title": "Toward Vision Language Model-based Assessment of Clinical Quality and Usability of LGE-MR Images for Cardiac Ablation Planning", "authors": ["Bipasha Kundu", "Abhishek Chaturvedi", "Axel W. E. Wismueller", "Richard Simon", "Cristian A. Linte"], "year": 2026, "published": "2026-08-21T14:51:48Z", "updated": "2026-08-21T14:51:48Z", "arxiv_id": "http://arxiv.org/abs/2608.21180v1", "url": "https://arxiv.org/abs/2608.21180v1", "primary_category": "eess.IV", "categories": ["eess.IV", "cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Interpretability", "Safety / Jailbreak", "VLM"], "benchmarks": [], "models": ["DeepSeek", "InternVL2"], "key_results": ["We curate a dataset of 60 annotated image slice-text pairs from 20 patients and benchmark four state-of-the-art VLM architectures.", "InternVL2 achieves the highest criterion-level accuracy (Avg ACC=0.65, PLCC=0.79), while DeepSeek achieves perfect clinical usability agreement (Acc=1.00, kappa=1.00)."], "abstract": "LGE cardiac MRI is widely used for left atrial fibrosis assessment and ablation planning in atrial fibrillation patients as knowledge of fibrotic tissue regions identified from LGE-MRI is critical for catheter ablation. Often, poor quality images used during ablation planning can cause mis-localization of ablation targets, directly impacting procedure safety and outcome. The decision of whether a scan meets the minimum quality threshold for ablation planning is currently made informally by the reviewing radiologist and is not captured by any automated system, yet it is arguably the most safety-critical output of the image quality assessment (IQA) process. However, variations in image quality caused by noise, motion artifacts, and poor boundary definition significantly compromise the reliability of downstream segmentation and clinical decision-making tasks. Manual quality assessment by expert radiologists is subjective and difficult to scale, while existing automated methods produce scalar scores without interpretable clinical reasoning. In this work, we propose a two-stage vision language model (VLM) framework for clinically grounded image quality assessment of left atrial LGE-MRI. In the first stage, a fine-tuned VLM generates structured radiology-style quality reports predicting five radiologist-defined criteria: Noise, Motion Artifact, LA Boundary Accuracy, PV Region Accuracy, and Under-segmentation Severity. In the second stage, a GPT-based reasoning module maps the predicted quality and reports to a structured quality scores and binary clinical usability decision for ablation planning. We curate a dataset of 60 annotated image slice-text pairs from 20 patients and benchmark four state-of-the-art VLM architectures. InternVL2 achieves the highest criterion-level accuracy (Avg ACC=0.65, PLCC=0.79), while DeepSeek achieves perfect clinical usability agreement (Acc=1.00, kappa=1.00).", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-03826fa227867ca7", "title": "Is Visual Prompting All You Need? Studying VLM Spatial Reasoning under Progressive Visual Scaffolds", "authors": ["Lars Benedikt Kaesberg", "Tianyu Yang", "Florian Valentin Wunderlich", "Terry Ruas", "Jan Philip Wahle", "Daniel Kurzawe", "Bela Gipp"], "year": 2026, "published": "2026-08-21T14:40:26Z", "updated": "2026-08-21T14:40:26Z", "arxiv_id": "http://arxiv.org/abs/2608.21170v1", "url": "https://arxiv.org/abs/2608.21170v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.AI"], "pillar": "Multimodal / World Models", "methods": ["RLVR / GRPO", "VLM"], "benchmarks": [], "models": ["ying", "yielding"], "key_results": ["Across multiple VLMs, these scaffolds improve task accuracy over the original visual setting by up to 34.0 percentage points and further complement GRPO-based training, yielding up to 4.6 additional accuracy points compared with near-zero gains on the original visual input.", "Analyses on both end-to-end task solving and object detection show that these gains are closely tied to reductions in grounding-related errors, while rule reasoning remains comparatively challenging."], "abstract": "Vision-language models (VLMs) have advanced rapidly in multimodal reasoning, yet recent work shows that their failures often reflect an interaction between visual grounding and downstream reasoning. What remains less clear is how the visual presentation of a task shapes model performance and failure modes when the underlying reasoning problem is unchanged. We study this question in SPaRC, a benchmark for grid-based visual spatial planning, by introducing lightweight input-side scaffolds that preserve the visual modality while making spatial structure more accessible. Across multiple VLMs, these scaffolds improve task accuracy over the original visual setting by up to 34.0 percentage points and further complement GRPO-based training, yielding up to 4.6 additional accuracy points compared with near-zero gains on the original visual input. Analyses on both end-to-end task solving and object detection show that these gains are closely tied to reductions in grounding-related errors, while rule reasoning remains comparatively challenging. We find that visual presentation is a central factor that determines whether VLM benchmarks measure grounded perception, downstream reasoning, or a mixture of both.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-f87b2e336d82c5d0", "title": "Masking Is Not Enough: Generative Restoration for Multimodal De-Identification in Medical AI", "authors": ["Shiva Shrestha", "Zongxing Xie", "Chen Zhao", "Liran Ma", "Zhipeng Cai", "Honghui Xu"], "year": 2026, "published": "2026-08-21T14:08:45Z", "updated": "2026-08-21T14:08:45Z", "arxiv_id": "http://arxiv.org/abs/2608.21133v1", "url": "https://arxiv.org/abs/2608.21133v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.CR"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": ["PHI", "PHI.", "ying"], "key_results": ["Medical image-text data can expose protected health information (PHI) through both visible image content as well as accompanying text, creating a barrier to privacy-preserving medical AI systems."], "abstract": "Medical image-text data can expose protected health information (PHI) through both visible image content as well as accompanying text, creating a barrier to privacy-preserving medical AI systems. This risk is especially prominent in multimodal systems, where images, questions, reports, and clinical context may enter training, evaluation, or inference pipelines. Existing medical vision-language benchmarks primarily emphasize task utility, while de-identification methods are often evaluated separately from downstream reasoning. We introduce ClinX, an end-to-end multimodal PHI sanitization framework for medical image-text data. ClinX detects visible identifiers with optical character recognition (OCR), constructs binary PHI masks, and applies ClinX-PRISM, a no-skip generative restoration module with privacy-oriented post-processing for burned-in identifier suppression. In parallel, text-side PHI is reduced through progressive de-identification levels: regex masking, context-aware masking, and rewrite-based sanitization. We evaluate ClinX in medical visual question answering (MedVQA), jointly measuring PHI leakage and downstream utility across image-side, text-side, and combined de-identification settings. Results show that OCR-only masking is not sufficient as a standalone solution, and restoration-based sanitization better preserves clinically relevant visual context while sharply reducing recoverable PHI.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-424dbb5927d9e7ee", "title": "AudioWorldSim: Realistic Binaural Audio Datasets For World Models", "authors": ["Luis Vitor Zerkowski", "Luiz Velho"], "year": 2026, "published": "2026-08-21T13:17:06Z", "updated": "2026-08-21T13:17:06Z", "arxiv_id": "http://arxiv.org/abs/2608.21075v1", "url": "https://arxiv.org/abs/2608.21075v1", "primary_category": "cs.SD", "categories": ["cs.SD", "cs.LG"], "pillar": "Multimodal / World Models", "methods": ["World Model"], "benchmarks": [], "models": [], "key_results": ["This technical report presents AudioWorldSim, an open-source platform designed to generate realistic binaural audio datasets and advance research in audio-based machine learning, particularly world models."], "abstract": "This technical report presents AudioWorldSim, an open-source platform designed to generate realistic binaural audio datasets and advance research in audio-based machine learning, particularly world models. Built as a custom extension of Meta's SoundSpaces 2.0 platform, AudioWorldSim leverages their comprehensive acoustics framework, but focuses on the automatic rollout of random agent navigations, as well as implements crucial fixes to how continuous sound is composed. AudioWorldSim is made publicly available to the research community at https://github.com/Luizerko/AudioWorldSim to facilitate reproducibility.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-92fe7246d43ec7b2", "title": "Recognition-Conditioned Reasoning: A Training-Free Multimodal-LLM Pipeline for Fine-Grained Micro-Action Understanding", "authors": ["Fengshun Wang", "Jin'ang Han", "Zhigang Tu"], "year": 2026, "published": "2026-08-21T12:10:32Z", "updated": "2026-08-21T12:10:32Z", "arxiv_id": "http://arxiv.org/abs/2608.21022v1", "url": "https://arxiv.org/abs/2608.21022v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.MM"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["This architecture achieves a statistically significant performance advantage on open-ended tasks, attaining an average score of 2.68 (on a five-point scale) compared to 1.44 for the second-best approach."], "abstract": "Micro-actions are subtle, short, low-amplitude body movements, such as a fidgeting hand or a slight head tilt, that humans perform with little conscious intent yet that reliably leak emotional and psychological state. Understanding them goes beyond assigning a label: a model must also describe which body parts move and reason, faithfully, about why a clip warrants a particular fine-grained category. We present the training-free, prompt-only system that won first place in the fine-grained understanding track (MA-Bench) of the MAC~2026 Micro-Action Challenge, where both fine-tuning and ground-truth supervision are disallowed. Built entirely upon frozen multimodal large language models (MLLMs), the system dynamically routes each of the eight sub-tasks to the MLLM empirically best suited for that task: a discriminative MLLM for closed-ended recognition tasks and a generative MLLM for open-ended description and reasoning tasks. This architecture achieves a statistically significant performance advantage on open-ended tasks, attaining an average score of 2.68 (on a five-point scale) compared to 1.44 for the second-best approach.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-c308c5ac6bcfcda1", "title": "Latent Ordinal Evidence, Misaligned Outputs: Inference-Time Ordinal Lens Alignment for Multimodal LLMs", "authors": ["Haiming Li", "Yingsheng Liu", "Jingmin Zhu", "Siyuan Yan", "Xieji Li", "Jiajun Sun", "Zhen Yu", "Zongyuan Ge"], "year": 2026, "published": "2026-08-21T11:33:22Z", "updated": "2026-08-21T11:33:22Z", "arxiv_id": "http://arxiv.org/abs/2608.20999v1", "url": "https://arxiv.org/abs/2608.20999v1", "primary_category": "cs.CV", "categories": ["cs.CV"], "pillar": "Multimodal / World Models", "methods": ["Test-time Scaling", "Preference Optimization", "VLM"], "benchmarks": [], "models": [], "key_results": ["OLA outperforms the SOTA LoRA-tuned OrderChain baseline in most settings while keeping the MLLM frozen, surpasses discriminative ordinal baselines in most cells, and improves over an offline lens in every setting."], "abstract": "Multimodal LLMs apply the language model interface to visual inputs, where ordinal regression tasks such as age estimation, image quality assessment, and disease grading require autoregressive decisions over ordered class labels. We ask whether MLLMs reliably convert internal ordinal evidence into ordered digit-token outputs. Across four ordinal benchmarks and four MLLM backbones, ordinal labels are linearly recoverable from hidden states with Spearman correlation up to 0.938, and a task-designed prompt further sharpens this structure. Yet native digit-token outputs weakly expose it: the unembedding matrix filters the ordinal direction, and the digit-token row space retains below 1.15% across all 16 model-dataset combinations, with a 16 to 77 absolute-point accuracy gap between linear-probe and native outputs. We introduce Ordinal Lens Alignment (OLA), a frozen-backbone inference-time method that trains lightweight W_S-anchored lenses on mid-to-deep decoder layers, fuses them into an ordinal distribution, and corrects only digit-token logits at generation. OLA outperforms the SOTA LoRA-tuned OrderChain baseline in most settings while keeping the MLLM frozen, surpasses discriminative ordinal baselines in most cells, and improves over an offline lens in every setting.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-2ed7bff8d85ecb47", "title": "MigrationNarrate: A Dataset for Detection of Migration Narratives in YouTube Videos", "authors": ["Fatima Haouari", "Carolina Scarton", "Kalina Bontcheva"], "year": 2026, "published": "2026-08-21T11:10:51Z", "updated": "2026-08-21T11:10:51Z", "arxiv_id": "http://arxiv.org/abs/2608.20984v1", "url": "https://arxiv.org/abs/2608.20984v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.CL", "cs.CY"], "pillar": "Multimodal / World Models", "methods": ["VLM"], "benchmarks": [], "models": [], "key_results": ["Narratives are central to how social communication is framed, making their detection critical for understanding and analysing public discourse."], "abstract": "Narratives are central to how social communication is framed, making their detection critical for understanding and analysing public discourse. Prior work has explored narrative detection and extraction across diverse domains; however, migration narratives remain significantly understudied, primarily due to the absence of dedicated annotated datasets. Furthermore, public communication has recently shifted towards video-centric platforms, where narratives are conveyed through multimodal signals and consumed at scale. Despite this shift, narratives in videos remain largely unexplored. To bridge these gaps, we introduce MigrationNarrate, the first multimodal dataset for detection of migration narratives in the UK, consisting of 1,115 YouTube video transcripts annotated using a two-level taxonomy of 12 migration super-narratives and 53 narrative labels. This paper details the dataset design, collection, and annotations; together with benchmark results using a combination of pre-trained encoder models and both open- and closed-source Large Language Models. Finally, a thorough error analysis offers insights for future work.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-d43ae07767716f0d", "title": "On the Transferability of Agricultural Weed Detection Under Cross-Field Distribution Shift", "authors": ["Nikhilesh Prabhakar", "Pranuthi Tenali", "Wilfredo Abudeye Fernandez", "Shekhar Borah", "Athresh Karanam", "Erik Blasch", "Prabha Sundaravadivel", "Sriraam Natarajan"], "year": 2026, "published": "2026-08-21T16:06:09Z", "updated": "2026-08-21T16:06:09Z", "arxiv_id": "http://arxiv.org/abs/2608.21254v1", "url": "https://arxiv.org/abs/2608.21254v1", "primary_category": "cs.CV", "categories": ["cs.CV", "cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": ["phistication", "yield"], "key_results": ["Using these datasets, we evaluate the performance of several strategies for transferring a detector trained on one crop to another, comparing unsupervised domain adaptive object detection (DAOD) against pretraining on a domain-adjacent source dataset followed by few-shot fine-tuning on the target dataset.", "We find that few-shot fine-tuning with as few as 25 labeled target examples outperforms unsupervised DAOD in our cross-crop comparison, suggesting that source domain selection combined with modest target supervision is more productive than algorithmic sophistication in adaptation."], "abstract": "Accurate agricultural weed detection in real-world field conditions is essential for precision agriculture, enabling targeted intervention and reducing yield loss. Recent work has reported strong detection performance from UAV-based imagery across a range of crops, yet existing approaches evaluate within a single crop and field, leaving practitioners with little evidence that a model trained on one crop will generalize to a new field or crop type. In this work, we characterize where cross-dataset weed-localization performance degrades and which modeling choices recover it, reducing the need to relabel every new deployment field. We introduce a newly collected and annotated UAV image dataset for agricultural weed detection in cotton fields and use it alongside an existing soybean dataset collected under a similar protocol. Using these datasets, we evaluate the performance of several strategies for transferring a detector trained on one crop to another, comparing unsupervised domain adaptive object detection (DAOD) against pretraining on a domain-adjacent source dataset followed by few-shot fine-tuning on the target dataset. Our analysis spans target-domain label budgets from zero to the full target dataset, characterizing the trade-off between adaptation strategy and annotation effort. We find that few-shot fine-tuning with as few as 25 labeled target examples outperforms unsupervised DAOD in our cross-crop comparison, suggesting that source domain selection combined with modest target supervision is more productive than algorithmic sophistication in adaptation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-003bd6a389497f05", "title": "Decoupling Policy Extraction for Offline Reinforcement Learning", "authors": ["Xuyao Lin", "Yixiang Shan", "Jinru Duan", "Tao Yang", "Xinyu Zhao", "Runyu Lei", "Yiming Zhao", "Jiaxin Fan", "Zongbao Feng", "Peng Jia"], "year": 2026, "published": "2026-08-21T09:26:53Z", "updated": "2026-08-21T09:26:53Z", "arxiv_id": "http://arxiv.org/abs/2608.20909v1", "url": "https://arxiv.org/abs/2608.20909v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.RO"], "pillar": "Quant × AI", "methods": ["RLVR / GRPO"], "benchmarks": [], "models": [], "key_results": ["This coupled learning process is well motivated in online RL, where an improved actor collects new data that can further update the actor and the critic.", "However, training data remains fixed in offline RL, making actor-side policy improvement unable to generate new data to validate or correct the critic.", "Motivated by this observation, we revisit the conventional offline RL paradigm and propose decoupling policy improvement from actor training."], "abstract": "Offline RL methods commonly jointly train the actor and critic, where the critic is used to guide the actor toward higher-value actions. This coupled learning process is well motivated in online RL, where an improved actor collects new data that can further update the actor and the critic. However, training data remains fixed in offline RL, making actor-side policy improvement unable to generate new data to validate or correct the critic. Moreover, retaining this coupled paradigm leads to two related challenges. Firstly, actor updates can drift toward high-valued but potentially out-of-distribution (OOD) actions and amplify critic overestimation. Secondly, conservative value estimation or behavior-cloning regularization creates a difficult trade-off between suppressing OOD actions and selecting high-value actions within the data-supported region. Motivated by this observation, we revisit the conventional offline RL paradigm and propose decoupling policy improvement from actor training. Specifically, we train the actor solely to model the behavior distribution and perform policy improvement at inference time by reranking multiple actor-generated proposals with a separately learned critic. We refer to this paradigm as the decoupled policy extraction paradigm. Under such paradigm, the actor provides behavior-supported action candidates, while the critic performs value-based selection within this candidate set. Extensive experiments show that the decoupled policy extraction paradigm outperforms both behavior cloning and jointly learned offline RL methods, while remaining effective even with a naive Q-learning critic.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-ed673ba6cb41f8da", "title": "Calibrating Inelastic Markets to Options: The Lean Marketron and the Generalized Langevin Equation", "authors": ["Andrey Itkin"], "year": 2026, "published": "2026-08-20T22:00:09Z", "updated": "2026-08-20T22:00:09Z", "arxiv_id": "http://arxiv.org/abs/2608.20589v1", "url": "https://arxiv.org/abs/2608.20589v1", "primary_category": "q-fin.CP", "categories": ["q-fin.CP", "physics.comp-ph", "q-fin.MF", "q-fin.PR"], "pillar": "Quant × AI", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": ["yields"], "key_results": ["The Marketron model of \\cite{HalperinItkin2025Mark} and its option pricing extension in \\cite{HalperinItkinMarketron2} suffer from structural non-identifiability: an eighteen-parameter space traps solvers in suboptimal local minima and renders economic quantities unmeasurable."], "abstract": "The Marketron model of \\cite{HalperinItkin2025Mark} and its option pricing extension in \\cite{HalperinItkinMarketron2} suffer from structural non-identifiability: an eighteen-parameter space traps solvers in suboptimal local minima and renders economic quantities unmeasurable. By removing exact scaling gauges and sign symmetries, freezing non-financial parameters by explicit criteria, and adiabatically eliminating the fast hidden signal, we derive a robust nine-parameter reduced model. A Gauss-Newton Hessian with empty null space and a manifold-boundary analysis confirm that the reduced core carries no exact symmetry and admits no further reduction. A diffusive correlation between flow and return innovations captures the short-maturity skew. A staged calibration from the physical measure to the risk-neutral measure, illustrated on SPX options, fits the whole surface with a single parameter set. The same reduction turns the wedge between the physical and pricing values of the flow block into a well-defined market price of flow risk rather than a ridge artifact, identifiable here for the first time, though a single surface constrains its level only weakly. Finally, our analysis reveals that in the Marketron model the log-price obeys a generalized Langevin equation with a closed-form, state-modulated memory kernel, and that the memory variable itself is the exact Markovian lift of this kernel. This mapping also yields a testable condition, the equality of the signal and memory relaxation rates, which on the SPX surface come out well separated, though both weakly identified, placing the fitted market tentatively in the driven, non-equilibrium regime and turning the active-matter reading from an analogy into a falsifiable constraint.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-6c3af1b9450fa6ef", "title": "FlavourBench: Ranking Frontier Language Models with Executable Culinary Ground Truth", "authors": ["Josef Chen", "Erim Hayretci"], "year": 2026, "published": "2026-08-20T21:14:52Z", "updated": "2026-08-20T21:14:52Z", "arxiv_id": "http://arxiv.org/abs/2608.20574v1", "url": "https://arxiv.org/abs/2608.20574v1", "primary_category": "cs.AI", "categories": ["cs.AI", "cs.CY", "cs.LG", "cs.SE"], "pillar": "Quant × AI", "methods": ["Verifier / PRM", "Quant / Trading"], "benchmarks": [], "models": ["Grok 4.6 has the largest point estimate at 65.1"], "key_results": ["Open-ended language-model benchmarks usually inherit a judge: a human preference panel, another model, or a brittle exact-match key."], "abstract": "Open-ended language-model benchmarks usually inherit a judge: a human preference panel, another model, or a brittle exact-match key. We introduce FlavourBench, an automated benchmark in which a versioned culinary system supplies dense, executable ground truth. Each task presents eight ingredients and asks for a three-ingredient portfolio; before model execution, Epicure scores all 56 possible portfolios. We evaluate 27 frontier endpoints on an identical 534-task core spanning substitution, pairing, and constrained composition. Every ranked model has exactly 89 valid responses per panel and family (14,418 model-task cells total), eliminating differential missingness from the leaderboard. The FlavourBench Score is the equal-family mean of the frozen task scores. We use 50,000 anchor-cluster bootstrap replicates for simultaneous 95% score bands and 100,000 sign-flip draws for all 351 paired model contrasts, with Holm control. The two independently compiled panels correlate at r = 0.89 (rank rho = 0.80). Grok 4.6 has the largest point estimate at 65.1 (simultaneous 95% CI 61.0-69.2); 101 of 351 model pairs are resolved. The release includes the prompts, all portfolio score maps, raw responses, exact routes, content hashes, and an offline verifier that reconstructs every result.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-1964f32b90f3f585", "title": "aiXamine: Unified Black-Box Evaluation of Cross-Dimensional Trade-offs in LLM Safety, Security, and Privacy", "authors": ["Fatih Deniz", "Yazan Boshmaf", "Dorde Popovic", "Issa Khalil"], "year": 2026, "published": "2026-08-20T20:33:35Z", "updated": "2026-08-20T20:33:35Z", "arxiv_id": "http://arxiv.org/abs/2608.20554v1", "url": "https://arxiv.org/abs/2608.20554v1", "primary_category": "cs.CR", "categories": ["cs.CR", "cs.LG"], "pillar": "Quant × AI", "methods": ["Distillation", "Preference Optimization", "Safety / Jailbreak"], "benchmarks": [], "models": ["phically", "ying"], "key_results": ["The critical failure modes in deployed large language models (LLMs) are cross-dimensional: a model can score 99.3 in safety alignment while refusing one in three benign queries, or improve across every capability metric while losing 21 points in privacy."], "abstract": "The critical failure modes in deployed large language models (LLMs) are cross-dimensional: a model can score 99.3 in safety alignment while refusing one in three benign queries, or improve across every capability metric while losing 21 points in privacy. Existing evaluation frameworks that assess safety, security, and privacy independently cannot detect these patterns. We introduce aiXamine, a unified black-box platform that evaluates LLM trustworthiness across safety, security, and privacy as interdependent properties. aiXamine orchestrates 46 tests across nine services through an automated red-teaming pipeline, producing hierarchical risk profiles, from prompt-level diagnostics to cross-service trade-off analytics, that enable reproducible comparison of proprietary and open-weight systems under identical conditions. Applying aiXamine to over 120 LLMs through more than 5,000 test runs, we conduct the largest joint safety, security, and privacy study to date and uncover three cross-dimensional phenomena invisible to single-axis evaluation. First, safety enforcement incurs a quantifiable safety tax: stronger alignment systematically increases over-refusal, forcing providers to choose between protection and utility. Second, privacy is near-orthogonal to other trustworthiness dimensions and not captured by standard alignment. Third, we identify and formally characterize distillation-induced robustness collapse: off-policy distillation without on-policy correction causes entropy collapse, catastrophically destroying robustness (56.9$\\to$2.6) on the same base architecture. These findings, compounded by diminishing returns from scale and category-dependent safety behaviors, demonstrate that trustworthiness is inherently multi-dimensional: progress along one axis does not guarantee, and can actively undermine, progress along others, yet current alignment methods treat it as a single objective.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-cd800a1395b51a07", "title": "A comparison between ceiling-mounted FMCW, IR-UWB and Wi-Fi radar for in-bedroom human activity monitoring and sleep interruption detection", "authors": ["Anton Lambrecht", "Reda El Hail", "Xianjun Jiao", "Pieter Crombez", "Dominique Schreurs", "Peter Karsmakers", "Adnan Shahid", "Eli De Poorter"], "year": 2026, "published": "2026-08-20T17:58:22Z", "updated": "2026-08-20T17:58:22Z", "arxiv_id": "http://arxiv.org/abs/2608.20322v1", "url": "https://arxiv.org/abs/2608.20322v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": [], "benchmarks": [], "models": [], "key_results": ["IR-UWB achieves the highest cross-subject activity recognition performance (89.0% macro F1), while FMCW generalizes best to unseen room layouts (83.8% macro F1)."], "abstract": "Despite their growing importance for contact-free radio frequency (RF) based healthcare monitoring, different radio technologies such as frequency-modulated continuous wave (FMCW) radar, impulse radio ultra-wideband (IR-UWB), and Wi-Fi sensing are rarely compared under identical deployment conditions, as existing studies typically differ in hardware, datasets, and evaluation methodologies. In addition, the performance of ceiling-mounted radars, despite their practical deployment and cost advantages in healthcare environments, remain underexplored. Therefore, this paper presents a controlled comparison and analysis of ceiling-mounted FMCW, IR-UWB, and Wi-Fi sensing using synchronized recordings from 20 participants across six room layouts. All technologies are evaluated with the same convolutional neural network (CNN) on both a fine-grained 10-class human activity recognition (HAR) task and a coarse 4-class sleep monitoring task. IR-UWB achieves the highest cross-subject activity recognition performance (89.0% macro F1), while FMCW generalizes best to unseen room layouts (83.8% macro F1). For sleep monitoring, all technologies exceed 92% macro F1 in unseen environments. The results reveal a fundamental trade-off between recognition performance and environmental robustness, which can be explained through differences in range resolution, antenna diversity, Doppler resolution, and spatial information retention. These findings provide practical guidelines for the design of healthcare-oriented RF sensing systems.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-930417586ba7fced", "title": "The Reconfiguration Premium: Co-movement Structure as an Unspanned Dimension of the Variance Risk Premium", "authors": ["Lucas Carvalho"], "year": 2026, "published": "2026-08-20T13:31:24Z", "updated": "2026-08-20T13:31:24Z", "arxiv_id": "http://arxiv.org/abs/2608.20020v1", "url": "https://arxiv.org/abs/2608.20020v1", "primary_category": "q-fin.MF", "categories": ["q-fin.MF", "q-fin.PR", "q-fin.ST"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": [], "key_results": ["Hedge ratios, factor models and diversified portfolios all rest on an estimate of which firms move together."], "abstract": "Hedge ratios, factor models and diversified portfolios all rest on an estimate of which firms move together. That estimate is not stable: firms migrate between the groupings the market treats as coherent, and when enough migrate the organizing axes of the cross-section turn. We measure the rate of that turning as the mean squared sine of the principal angles between subdominant eigenspaces of consecutive twelve-month S&P 500 correlation matrices. A typical month rewrites a fifth of the structure and carries four-fifths forward. That rate is priced: it couples to the aggregate variance risk premium at t = 5.40, no level measure correlates above 0.32, and the implied-correlation surface spans at most 6.7 percent of it. Only the persistent component is priced - the premium compensates the pace of revision, not the distance traveled. The mechanism is prepayment: implied variance rises on impact, volatility follows two to three quarters later (simulated-null p < 0.03 at h = 1-9), and the premium converges as it arrives. Three pre-registered boundaries hold: no timing alpha, no crash protection, and a downside version inseparable from intensity. The premium is, in part, rent on exposure held over a map still being redrawn.", "code_available": true, "confidence": 0.8, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-9df9f0d0f32e3da8", "title": "Interpretable Feature Learning for RF Fingerprinting via Polar MKANs", "authors": ["Mikhail Krasnov", "Ljupcho Milosheski", "Carolina Fortuna"], "year": 2026, "published": "2026-08-20T10:42:33Z", "updated": "2026-08-20T10:42:33Z", "arxiv_id": "http://arxiv.org/abs/2608.19881v1", "url": "https://arxiv.org/abs/2608.19881v1", "primary_category": "eess.SP", "categories": ["eess.SP", "cs.LG"], "pillar": "Quant × AI", "methods": ["Interpretability"], "benchmarks": [], "models": ["yielding"], "key_results": ["Radio frequency (RF) fingerprinting authenticates wireless devices from hardware-induced I/Q impairments, typically with deep learning feature extractors that are accurate but opaque, limiting their use in security critical settings."], "abstract": "Radio frequency (RF) fingerprinting authenticates wireless devices from hardware-induced I/Q impairments, typically with deep learning feature extractors that are accurate but opaque, limiting their use in security critical settings. We propose Polar Monotonic Kolmogorov-Arnold Networks (Polar MKAN), a block partitioned monotonic encoder on polar inputs in which each latent dimension depends exclusively on magnitude or phase, yielding channel separation and monotone responses by construction. On a synthetic gain and carrier frequency offset (CFO) benchmark, Polar MKAN reaches 57.2 percent DCI Disentanglement versus at most 12.9 percent for unpartitioned baselines. We further evaluate the detection accuracy trade off on real data and the sensitivity to blind CFO compensation.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-cc8aed71a81e5991", "title": "Inadvertent Context Leakage in Language Models", "authors": ["Jaiden Fairoze", "Neal Mangaokar", "Kamalika Chaudhuri", "Sanjam Garg", "Saeed Mahloujifar"], "year": 2026, "published": "2026-08-20T10:05:29Z", "updated": "2026-08-20T10:05:29Z", "arxiv_id": "http://arxiv.org/abs/2608.19857v1", "url": "https://arxiv.org/abs/2608.19857v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.CR"], "pillar": "Quant × AI", "methods": ["Long Context", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["For AI agents to be useful beyond simple chat, they must hold sensitive user context such as calendars, credentials, health records, and financial data."], "abstract": "For AI agents to be useful beyond simple chat, they must hold sensitive user context such as calendars, credentials, health records, and financial data. We study whether the mere presence of such secrets in a model's context window introduces hidden correlations into the model's benign outputs, allowing reconstruction even when the model correctly refuses direct extraction. We further study whether an adversary can actively engineer prompts that amplify this effect, using the model as a covert carrier to transmit secrets through seemingly innocuous text. In both cases, this limited leakage is exploited using a novel adaptive attack that assumes black-box access to the underlying model.\n  In controlled experiments across eight proprietary models, we find that 2-digit in-context secrets are reconstructed with near-perfect accuracy and 4-digit secrets at 82\\% exact match, all from outputs the model produces in response to ordinary, non-adversarial requests. We observe that more capable models leak more: stronger instruction-following amplifies sensitivity to in-context secrets, suggesting leakage is a byproduct of capability as opposed to a patchable bug. We show this leakage enables two practical attacks: (1) a trained classifier that infers semantic predicates about user memories (e.g., health conditions, financial events) from routine natural-language outputs, and (2) an RL-trained adversary that extracts full Social Security Numbers from a production-style agent.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-2c40424f7c7fc559", "title": "FAR-DPO: Feasibility-Aware and Robust Direct Preference Optimization for Cyclic Peptide Design", "authors": ["Guofeng Zhang", "Rong Han", "Xiaoyu Wang", "Zhiyun Li", "Zongbo Han", "Xiaohong Liu", "Guangyu Wang"], "year": 2026, "published": "2026-08-20T09:00:26Z", "updated": "2026-08-20T09:00:26Z", "arxiv_id": "http://arxiv.org/abs/2608.19808v1", "url": "https://arxiv.org/abs/2608.19808v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Preference Optimization"], "benchmarks": [], "models": ["yields"], "key_results": ["These gains also extend to the hardest target quartile and are accompanied by more favorable best-per-target binding scores."], "abstract": "Cyclic peptides are emerging as promising molecular scaffolds in drug discovery due to their high binding affinity and structural stability. However, extending generative models from linear to cyclic peptide design remains challenging, as cyclization sharply restricts the feasible design space through coupled geometric and biophysical constraints. Moreover, limited training data has led existing approaches to rely largely on zero-shot generation or post hoc filtering, resulting in low yields of feasible designs and limited control over multi-objective trade-offs. To address these limitations, we propose FAR-DPO (Feasibility-Aware and Robust Direct Preference Optimization), an architecture-agnostic framework that steers generative models toward structurally and biophysically feasible cyclic peptide designs, particularly for challenging targets. FAR-DPO integrates feasibility-aware preference construction with difficulty-aware group-robust optimization. Specifically, it constructs within-target preference pairs through feasibility-gated multi-objective dominance and adaptively reweights predefined difficulty groups according to their current preference losses. On the CPSea LNR benchmark, under a fixed generation budget, FAR-DPO increases overall success rate from 46.89% to 57.79% on PepGLAD and from 47.96% to 49.57% on PepFlow. These gains also extend to the hardest target quartile and are accompanied by more favorable best-per-target binding scores. Together, these results demonstrate FAR-DPO's effectiveness in improving feasibility and target-wise robustness.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-5d7b3ac24bfd3d30", "title": "Unsupervised Anomaly Detection Using Flow Matching on Tabular Data", "authors": ["Philip Konz", "Tejaswini Medi", "Margret Keuper"], "year": 2026, "published": "2026-08-20T08:56:46Z", "updated": "2026-08-20T08:56:46Z", "arxiv_id": "http://arxiv.org/abs/2608.19801v1", "url": "https://arxiv.org/abs/2608.19801v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["With these scores, Forest-Flow becomes competitive with, and in some cases outperforms, TCCM."], "abstract": "Financial anomaly detection often relies on large unlabeled transaction logs, where anomalous samples may already be present during training. Such training-set contamination violates the clean-normal data assumption underlying many anomaly detection methods. Although flow matching has demonstrated strong performance in generative modeling, its robustness in unsupervised tabular anomaly detection remains underexplored. In this work, we study flow-matching-based anomaly detection under contaminated training data by comparing Time-Conditioned Contraction Matching (TCCM) with Forest-Flow and evaluating multiple anomaly scoring functions. Our results show that the choice of anomaly score is critical. The original single-step Decision score used by TCCM is sensitive to contamination, whereas trajectory-based Deviation and Reconstruction scores provide more stable anomaly signals. With these scores, Forest-Flow becomes competitive with, and in some cases outperforms, TCCM. These findings highlight the importance of anomaly scoring for flow-matching methods in financial anomaly detection under severe class imbalance.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-6b1f24951c2660cf", "title": "In Two Minds about Lifelong Learning: Exploring Hemispheric Redundancy and Specialisation in Neural Models", "authors": ["Benjamin Smith", "Levin Kuhlmann", "Kaushik Roy", "Gideon Kowadlo"], "year": 2026, "published": "2026-08-20T00:15:18Z", "updated": "2026-08-20T00:15:18Z", "arxiv_id": "http://arxiv.org/abs/2608.19514v1", "url": "https://arxiv.org/abs/2608.19514v1", "primary_category": "cs.LG", "categories": ["cs.LG", "cs.AI"], "pillar": "Quant × AI", "methods": ["Memory / RAG", "Quant / Trading"], "benchmarks": [], "models": ["phic"], "key_results": ["Finally, we present results showing that our architecture achieves competitive results on the Split-MNIST, Split-Fashion-MNIST and Split-CIFAR-100 datasets, with 98.3%, 84.9%, and 29.29% accuracy respectively."], "abstract": "Persistent intelligent systems require the ability to learn continually, but current machine learning approaches face significant challenges in this area compared to biological learning systems. Machine learning algorithms typically trade off retention of previously learned information and adaptation to new or changing data patterns. When continual learning capabilities are absent, algorithms must undergo retraining using the entire data set, an approach that becomes impractical when original training data are unavailable due to storage constraints, financial or computational costs, or privacy restrictions. However, biological animals can learn continually, without experiencing catastrophic forgetting. This paper attempts to build a high-level framework for how animals learn and preserve knowledge by modelling neural components and states that are known to be related to memory consolidation. We focus on three concepts: experience replay, REM sleep, and bilaterality. We propose 4MAS (4 Module Awake/Sleep), a novel macroarchitecture demonstrating how machine learning models might benefit from asymmetric hemispheres, each with their own long- and short-term memory mechanisms, and how a period of sleep between incremental learning tasks might benefit memory consolidation. Finally, we present results showing that our architecture achieves competitive results on the Split-MNIST, Split-Fashion-MNIST and Split-CIFAR-100 datasets, with 98.3%, 84.9%, and 29.29% accuracy respectively.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}
{"id": "evg-4c6f668e753625d8", "title": "Quantifying Event Impacts on Time Series via Multiscale Contrastive Learning", "authors": ["Yiming Sun", "Shengyu Chen", "Zhengzhang Chen", "Haoyu Wang", "Xiaowei Jia", "Haifeng Chen"], "year": 2026, "published": "2026-08-19T21:02:04Z", "updated": "2026-08-19T21:02:04Z", "arxiv_id": "http://arxiv.org/abs/2608.19447v1", "url": "https://arxiv.org/abs/2608.19447v1", "primary_category": "cs.LG", "categories": ["cs.LG"], "pillar": "Quant × AI", "methods": ["Interpretability", "Quant / Trading"], "benchmarks": [], "models": ["ying"], "key_results": ["Experiments show that EventTime consistently outperforms state-of-the-art time-series and event-aware baselines in estimating post-event financial losses."], "abstract": "Shocks that spread through the web, such as cybersecurity breach disclosures, can abruptly disrupt financial time series and cause substantial abnormal losses. While these events are disclosed as discrete records through news reports, regulatory filings, or public databases, their consequences unfold through continuous market dynamics. This creates an event-conditioned impact prediction problem: given pre-event market history and limited event metadata, the goal is to estimate short-term post-disclosure abnormal loss rather than reconstruct the full post-event trajectory. However, most time-series forecasting models focus on endogenous regularities such as trend, seasonality, and autocorrelation, and thus struggle with rare and heterogeneous external events. The challenge is further amplified by sparse high-impact events and background market noise. We introduce EventTime, a multi-resolution framework that combines long-horizon market context, short-horizon pre-event dynamics, and event metadata. It incorporates an event fusion module that couples temporal representations with event attributes to identify relevant recent market patterns. To mitigate sparse supervision, EventTime further introduces a dynamic contrastive objective that constructs event- and time-series-aware positive and negative pairs during training. We also construct SECURE, a real-world dataset aligning cybersecurity incidents with stock-market time series and structured and LLM-derived semantic features. Experiments show that EventTime consistently outperforms state-of-the-art time-series and event-aware baselines in estimating post-event financial losses. Further analyses demonstrate more event-sensitive representations, greater robustness to incomplete metadata, and more interpretable estimates of short-term market impact following cybersecurity disclosures.", "code_available": false, "confidence": 0.7, "swept_on": "2026-08-24T03:31:24.897183+00:00"}