# **GreatX**: Graph Reliability Toolbox
GreatX is great!
[Documentation]
|
[Examples]
# ❓ What is "Reliability" on Graphs?
"Reliability" on graphs refers to *robustness* against the following threats:
+ Inherent noise
+ Distribution Shift
+ Adversarial Attacks
For more details, please kindly refer to our paper [**Recent Advances in Reliable Deep Graph Learning: Inherent Noise, Distribution Shift, and Adversarial Attack**](https://arxiv.org/abs/2202.07114)
# 💨 News
- November 2, 2022: We are planning to release GreatX 0.1.0 this month, stay tuned!
- June 30, 2022: GraphWar has been renamed to GreatX.
- ~~June 9, 2022: GraphWar **v0.1.0** has been released. We also provide the [documentation](https://greatx.readthedocs.io/en/latest) along with numerous [examples](https://github.com/EdisonLeeeee/GreatX/blob/master/examples)~~ .
- ~~May 27, 2022: GraphWar has been refactored with [PyTorch Geometric (PyG)](https://github.com/pyg-team/pytorch_geometric), old code based on [DGL](https://www.dgl.ai) can be found [here](https://github.com/EdisonLeeeee/GreatX/tree/dgl). We will soon release the first version of GreatX, stay tuned!~~
NOTE: GreatX is still in the early stages and the API will likely continue to change.
If you are interested in this project, don't hesitate to contact me or make a PR directly.
# 🚀 Installation
Please make sure you have installed [PyTorch](https://pytorch.org) and [PyTorch Geometric (PyG)](https://pytorch-geometric.readthedocs.io/en/latest/notes/installation.html).
```bash
# Coming soon
pip install -U greatx
```
or
```bash
# Recommended
git clone https://github.com/EdisonLeeeee/GreatX.git && cd GreatX
pip install -e . --verbose
```
where `-e` means "editable" mode so you don't have to reinstall every time you make changes.
# ⚡ Get Started
Assume that you have a `torch_geometric.data.Data` instance `data` that describes your graph.
## How fast can we train and evaluate your own GNN?
Take `GCN` as an example:
```python
from greatx.nn.models import GCN
from greatx.training import Trainer
from torch_geometric.datasets import Planetoid
# Any PyG dataset is available!
dataset = Planetoid(root='.', name='Cora')
data = dataset[0]
model = GCN(dataset.num_features, dataset.num_classes)
trainer = Trainer(model, device='cuda:0') # or 'cpu'
trainer.fit(data, mask=data.train_mask)
trainer.evaluate(data, mask=data.test_mask)
```
## A simple targeted manipulation attack
```python
from greatx.attack.targeted import RandomAttack
attacker = RandomAttack(data)
attacker.attack(1, num_budgets=3) # attacking target node `1` with `3` edges
attacked_data = attacker.data()
edge_flips = attacker.edge_flips()
```
## A simple untargeted (non-targeted) manipulation attack
```python
from greatx.attack.untargeted import RandomAttack
attacker = RandomAttack(data)
attacker.attack(num_budgets=0.05) # attacking the graph with 5% edges perturbations
attacked_data = attacker.data()
edge_flips = attacker.edge_flips()
```
# 👀 Implementations
In detail, the following methods are currently implemented:
## ⚔ Adversarial Attack
### Graph Manipulation Attack (GMA)
#### Targeted Attack
| Methods | Descriptions | Examples |
| ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------- |
| **RandomAttack** | A simple random method that chooses edges to flip randomly. | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/random_attack.py) |
| **DICEAttack** | *Waniek et al.* [Hiding Individuals and Communities in a Social Network](https://arxiv.org/abs/1608.00375), *Nature Human Behavior'16* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/dice_attack.py) |
| **Nettack** | *Zügner et al.* [Adversarial Attacks on Neural Networks for Graph Data](https://arxiv.org/abs/1805.07984), *KDD'18* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/nettack.py) |
| **FGAttack** | *Chen et al.* [Fast Gradient Attack on Network Embedding](https://arxiv.org/abs/1809.02797), *arXiv'18* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/fg_attack.py) |
| **GFAttack** | *Chang et al*. [A Restricted Black - box Adversarial Framework Towards Attacking Graph Embedding Models](https://arxiv.org/abs/1908.01297), *AAAI'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/gf_attack.py) |
| **IGAttack** | *Wu et al.* [Adversarial Examples on Graph Data: Deep Insights into Attack and Defense](https://arxiv.org/abs/1903.01610), *IJCAI'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/ig_attack.py) |
| **SGAttack** | *Li et al.* [ Adversarial Attack on Large Scale Graph](https://arxiv.org/abs/2009.03488), *TKDE'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/sg_attack.py) |
| **PGDAttack** | *Xu et al.* [Topology Attack and Defense for Graph Neural Networks: An Optimization Perspective](https://arxiv.org/abs/1906.04214), *IJCAI'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/targeted/pgd_attack.py) |
#### Untargeted Attack
| Methods | Descriptions | Examples |
| ---------------- | ---------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- |
| **RandomAttack** | A simple random method that chooses edges to flip randomly | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/untargeted/random_attack.py) |
| **DICEAttack** | *Waniek et al.* [Hiding Individuals and Communities in a Social Network](https://arxiv.org/abs/1608.00375), *Nature Human Behavior'16* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/untargeted/dice_attack.py) |
| **FGAttack** | *Chen et al.* [Fast Gradient Attack on Network Embedding](https://arxiv.org/abs/1809.02797), *arXiv'18* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/untargeted/fg_attack.py) |
| **Metattack** | *Zügner et al.* [Adversarial Attacks on Graph Neural Networks via Meta Learning](https://arxiv.org/abs/1902.08412), *ICLR'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/untargeted/metattack.py) |
| **IGAttack** | *Wu et al.* [Adversarial Examples on Graph Data: Deep Insights into Attack and Defense](https://arxiv.org/abs/1903.01610), *IJCAI'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/untargeted/ig_attack.py) |
| **PGDAttack** | *Xu et al.* [Topology Attack and Defense for Graph Neural Networks: An Optimization Perspective](https://arxiv.org/abs/1906.04214), *IJCAI'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/untargeted/pgd_attack.py) |
### Graph Injection Attack (GIA)
| Methods | Descriptions | Examples |
| ------------------- | --------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| **RandomInjection** | A simple random method that chooses nodes to inject randomly. | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/injection/random_injection.py) |
| **AdvInjection** | The 2nd place solution of [KDD Cup 2020](https://www.biendata.xyz/competition/kddcup_2020/), team: ADVERSARIES. | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/injection/adv_injection.py) |
### Graph Universal Attack (GUA)
### Graph Backdoor Attack (GBA)
| Methods | Descriptions | Examples |
| --------------- | ------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------ |
| **LGCBackdoor** | *Chen et al.* [Neighboring Backdoor Attacks on Graph Convolutional Network](https://arxiv.org/abs/2201.06202), *arXiv'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/backdoor/lgc_backdoor.py) |
| **FGBackdoor** | *Chen et al.* [Neighboring Backdoor Attacks on Graph Convolutional Network](https://arxiv.org/abs/2201.06202), *arXiv'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/attack/backdoor/fg_backdoor.py) |
## Enhancing Techniques or Corresponding Defense
### Standard GNNs (without defense)
#### Supervised
| Methods | Descriptions | Examples |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------- |
| **GCN** | *Kipf et al.* [Semi-Supervised Classification with Graph Convolutional Networks](https://arxiv.org/abs/1609.02907), *ICLR'17* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/gcn.py) |
| **SGC** | *Wu et al.* [Simplifying Graph Convolutional Networks](https://arxiv.org/abs/1902.07153), *ICLR'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/sgc.py) |
| **GAT** | *Veličković et al.* [Graph Attention Networks](https://arxiv.org/abs/1710.10903), *ICLR'18* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/gat.py) |
| **DAGNN** | *Liu et al.* [Towards Deeper Graph Neural Networks](https://arxiv.org/abs/2007.09296), *KDD'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/dagnn.py) |
| **APPNP** | *Klicpera et al.* [Predict then Propagate: Graph Neural Networks meet Personalized PageRank](https://arxiv.org/abs/1810.05997), *ICLR'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/appnp.py) |
| **JKNet** | *Xu et al.* [Representation Learning on Graphs with Jumping Knowledge Networks](https://arxiv.org/abs/1806.03536), *ICML'18* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/jknet.py) |
| **TAGCN** | *Du et al.* [Topological Adaptive Graph Convolutional Networks](https://arxiv.org/abs/1806.03536), *arXiv'17* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/tagcn.py) |
| **SSGC** | *Zhu et al.* [Simple Spectral Graph Convolution](https://openreview.net/forum?id=CYO5T-YjWZV), *ICLR'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/ssgc.py) |
| **DGC** | *Wang et al.* [Dissecting the Diffusion Process in Linear Graph Convolutional Networks](https://arxiv.org/abs/2102.10739), *NeurIPS'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/dgc.py) |
| **NLGCN, NLMLP, NLGAT** | *Liu et al.* [Non-Local Graph Neural Networks](https://ieeexplore.ieee.org/document/9645300), *TPAMI'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/nlgnn.py) |
| **SpikingGCN** | *Zhu et al.* [Spiking Graph Convolutional Networks](https://arxiv.org/abs/2205.02767), *IJCAI'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/spiking_gcn.py) |
#### Unsupervised/Self-supervise
| Methods | Descriptions | Examples |
| ----------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- |
| **DGI** | *Veličković et al.* [Deep Graph Infomax](https://arxiv.org/abs/1809.10341), *ICLR'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/unsupervised/dgi.py) |
| **GRACE** | *Zhu et al.* [Deep Graph Contrastive Representation Learning](https://arxiv.org/abs/2006.04131), *ICML'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/unsupervised/grace.py) |
| **CCA-SSG** | *Zhang et al.* [From Canonical Correlation Analysis to Self-supervised Graph Neural Networks](https://arxiv.org/abs/2106.12484), *NeurIPS'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/unsupervised/cca_ssg.py) |
| **GGD** | *Zheng et al.* [Rethinking and Scaling Up Graph Contrastive Learning: An Extremely Efficient Approach with Group Discrimination](https://arxiv.org/abs/2206.01535), *NeurIPS'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/unsupervised/ggd.py) |
### Techniques Against Adversarial Attacks
| Methods | Descriptions | Examples |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| **MedianGCN** | *Chen et al.* [Understanding Structural Vulnerability in Graph Convolutional Networks](https://www.ijcai.org/proceedings/2021/310), *IJCAI'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/median_gcn.py) |
| **RobustGCN** | *Zhu et al.* [Robust Graph Convolutional Networks Against Adversarial Attacks](http://pengcui.thumedialab.com/papers/RGCN.pdf), *KDD'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/robust_gcn.py) |
| **SoftMedianGCN** | *Geisler et al.* [Reliable Graph Neural Networks via Robust Aggregation](https://arxiv.org/abs/2010.15651), *NeurIPS'20*
*Geisler et al.* [Robustness of Graph Neural Networks at Scale](https://arxiv.org/abs/2110.14038), *NeurIPS'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/soft_median_gcn.py) |
| **ElasticGNN** | *Liu et al.* [Elastic Graph Neural Networks](https://arxiv.org/abs/2107.06996), *ICML'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/elastic_gnn.py) |
| **AirGNN** | *Liu et al.* [Graph Neural Networks with Adaptive Residual](https://openreview.net/forum?id=hfkER_KJiNw), *NeurIPS'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/air_gnn.py) |
| **SimPGCN** | *Jin et al.* [Node Similarity Preserving Graph Convolutional Networks](https://arxiv.org/abs/2011.09643), *WSDM'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/simp_gcn.py) |
| **SAT** | *Li et al.* [Spectral Adversarial Training for Robust Graph Neural Network](https://arxiv.org/abs/2211.10896), *arXiv'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/model/supervised/gcn_sat.py) |
| **JaccardPurification** | *Wu et al.* [Adversarial Examples on Graph Data: Deep Insights into Attack and Defense](https://arxiv.org/abs/1903.01610), *IJCAI'19* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/defense/gcn_jaccard.py) |
| **SVDPurification** | *Entezari et al.* [All You Need Is Low (Rank): Defending Against Adversarial Attacks on Graphs](https://arxiv.org/abs/1903.01610), *WSDM'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/defense/gcn_svd.py) |
| **GNNGUARD** | *Zhang et al.* [GNNGUARD: Defending Graph Neural Networks against Adversarial Attacks](https://arxiv.org/abs/2006.08149), *NeurIPS'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/defense/gnnguard.py) |
| **GUARD** | *Li et al.* [GUARD: Graph Universal Adversarial Defense](https://arxiv.org/abs/2204.09803), *arXiv'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/defense/universal_defense.py) |
| **RTGCN** | *Wu et al.* [Robust Tensor Graph Convolutional Networks via T-SVD based Graph Augmentation](https://dl.acm.org/doi/abs/10.1145/3534678.3539436), *KDD'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/model/supervised/rt_gcn.py) |
More details of literatures and the official codes can be found at [Awesome Graph Adversarial Learning](https://github.com/gitgiter/Graph-Adversarial-Learning).
### Techniques Against Inherent Noise
| Methods | Descriptions | Examples |
| ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- |
| **DropEdge** | *Rong et al.* [DropEdge: Towards Deep Graph Convolutional Networks on Node Classification](https://arxiv.org/abs/1907.10903), *ICLR'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/drop_edge.py) |
| **DropNode** | *You et al.* [Graph Contrastive Learning with Augmentations](https://arxiv.org/abs/2010.13902), *NeurIPS'20* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/drop_node.py) |
| **DropPath** | *Li et al.* [MaskGAE: Masked Graph Modeling Meets Graph Autoencoders](https://arxiv.org/abs/2205.10053), *arXiv'22*' | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/models/supervised/drop_path.py) |
| **FeaturePropagation** | *Rossi et al.* [On the Unreasonable Effectiveness of Feature propagation in Learning on Graphs with Missing Node Features](https://openreview.net/forum?id=qe_qOarxjg), *Log'22* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/defense/feature_propagation.py) |
## Miscellaneous
| Methods | Descriptions | Examples |
| ----------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------- |
| **Centered Kernel Alignment (CKA)** | *Nguyen et al.* [Do Wide and Deep Networks Learn the Same Things? Uncovering How Neural Network Representations Vary with Width and Depth](https://arxiv.org/abs/2010.15327), *ICLR'21* | [[**Example**]](https://github.com/EdisonLeeeee/GreatX/blob/master/examples/cka.py) |
# ❓ Known Issues
+ `Untargeted attacks` are suffering from performance degradation, as also in DeepRobust, when a validation set is used during training with model picking. Such phenomenon has also been revealed in [Black-box Gradient Attack on Graph Neural Networks: Deeper Insights in Graph-based Attack and Defense](https://arxiv.org/abs/2104.15061).