{ "cells": [ { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "PmGDJzj1EAoJ" }, "source": [ "# Part 3 of 4 - Model Training and Analysis\n", "\n", "#### | Final Capstone Project for the Diploma Program in Data Science | BrainStation Vancouver |\n", "\n", "| Arash Tavassoli | May-June 2019 |\n", "\n", "---\n", "This is the third notebook in a series of four:\n", "* **Part 1 - Exploratory Data Analysis**\n", "\n", "* **Part 2 - Data Preprocessing**\n", "\n", "* **Part 3 - Model Training and Analysis**\n", "\n", "* **Part 4 - Real-Time Facial Expression Recognition**\n", "\n", "### What to expect in this notebook:\n", "\n", "1. A brief intoroduction to the approach taken in compiling and training the models\n", "2. A set of helper functions and an Excel spreadsheet that will be used for compiling and training the models\n", "3. Training of one model for each of the 3 and 5-expression classifiers\n", "3. Model performance analysis and comparison to human accuracies from Notebook 1\n", "\n", "---\n", "## 1. Convolutional Neural Networks:\n", "Convolutional Neural Networks (CNN) are currently considered as one of the top choices for any image classification problem, primarily because they are able to pick up on patterns in small parts of an image. However, this deep learning approach shows its true power only when it is presented with large enough data at the training stage. It is also considered as a computationally expensive Machine Learning system, given the large number of trainable parameters, specially in larger (deeper) network architectures.\n", "\n", "In this project we will use Python's high-level neural networks API, Keras (using TensorFlow backend) to build and tune a Convolutional Neural Network on the images that we processed in the previous notebooks.\n", "\n", "\n", "## 2. The Approach:\n", "\n", "Considering the large number of hyperparameters that can be tuned for a Convolutional Neural Network, the approach for this project is to build a set of helper functions that, given a model number, can read the hyperparameters for each convolutional (CONV) and fully connected (FC) layers, as well as the optimization parameters, from an Excel sheet that is saved as `Model Params.xlsx`. The functions will then compile the sequential model and save the trained model, the training history and some accuracy metrics on a new folder (with the same model number), and also update the Excel sheet with the training time, training accuracy and validation accuracy.\n", "\n", "Training will focus on a model with 3 and anoher model with 5 classes of facial expressions." ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "fkwQXhnCEAow" }, "source": [ "### 1.1. The Excel Sheet with Pre-Defined Hyperparameters\n", "\n", "The `Model Params.xlsx` is saved in the same folder as the current notebook and hosts the values for the following hyperparameters that will be used during model training:\n", "\n", "- **For each CONV layer:**\n", " - Number of filters (int)\n", " - Filter size (int)\n", " - Filter strides (int)\n", " - Filter padding type ('same' or 'valid')\n", " - Activation function ('relu', 'sigmoid', 'tanh')\n", " - Kernel Initializer ('Xavier')\n", " - L2 regularization weight (float)\n", " - Pooling type ('Average' or 'Max')\n", " - Pooling filter size (int)\n", " - Pooling filter strides (int)\n", " - Pooling filter padding type ('same' or 'valid') \n", " - Dropout rate (float between 0 and 1)\n", "\n", "\n", "- **For each FC layer:**\n", " - Number of filters (int)\n", " - Activation function ('relu', 'sigmoid', 'tanh')\n", " - L2 regularization weight (float)\n", " - Dropout rate (float between 0 and 1)\n", " \n", " \n", "- **Other parameters:**\n", " - Model number (int)\n", " - Number of classes (3 or 5)\n", " - Number of epochs (int)\n", " - Batch size (int)\n", " - Optimizer type ('Adam')\n", " - Learning rate (float)\n", " - Image input size (100x100 for this project)\n", " \n", "\n", "\n", "Let's start by importing the required libraries:" ] }, { "cell_type": "code", "execution_count": 1, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 34 }, "colab_type": "code", "executionInfo": { "elapsed": 2792, "status": "ok", "timestamp": 1560562211539, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "V6HnIrvtYWDN", "outputId": "b565c81c-813b-4eac-e7fa-1fd28e0245bc" }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "Using TensorFlow backend.\n" ] } ], "source": [ "import numpy as np\n", "import pandas as pd\n", "\n", "import matplotlib.pyplot as plt\n", "import matplotlib.gridspec as gridspec\n", "import seaborn as sns\n", "\n", "from sklearn.metrics import confusion_matrix\n", "from sklearn.metrics import classification_report \n", "\n", "import keras\n", "from keras.models import Sequential\n", "from keras.layers import Dense, Conv2D, Flatten, MaxPooling2D, Dropout, SpatialDropout2D, AveragePooling2D, BatchNormalization, Activation\n", "from keras import regularizers\n", "from keras.utils import plot_model\n", "from keras import models\n", "\n", "from google.colab import drive\n", "from IPython.display import clear_output\n", "import json\n", "import time\n", "import os" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "XfJLpwXdwVU-" }, "source": [ "The entire model training process has been done on Google Colab (to use GPUs), therefore the following root directory is specific to my personal Google Drive where the notebook was saved:" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "4zsNDj68EAoq" }, "outputs": [], "source": [ "# This root_dir is used after mounting the Google Drive on Google Colab environment\n", "root_dir = '/gdrive/My Drive/Colab Notebooks/BrainStation Capstone - Colab'" ] }, { "cell_type": "code", "execution_count": 4, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 34 }, "colab_type": "code", "executionInfo": { "elapsed": 232, "status": "ok", "timestamp": 1560562230570, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "F-CFcEfQYqYr", "outputId": "362d6310-5c02-411c-e94a-a55389ec875a" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Drive already mounted at /gdrive; to attempt to forcibly remount, call drive.mount(\"/gdrive\", force_remount=True).\n" ] } ], "source": [ "# Mount Google Drive\n", "drive.mount('/gdrive')" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "srvoOV2BwVVG" }, "source": [ "Let's load the Excel sheet, with hyperparameters already saved on it:" ] }, { "cell_type": "code", "execution_count": 5, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 337 }, "colab_type": "code", "executionInfo": { "elapsed": 1937, "status": "ok", "timestamp": 1560562237299, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "rULOqydiSwni", "outputId": "34922cbb-7731-4acf-c5ba-4a0d7a6001ef" }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
modellayertypesizefilter_sizefilter_sfilter_pactiv_funkernel_initL2_regpool_typepool_sizepool_spool_pdropout_ratenum_classesnum_epochsbatch_sizeoptimizerlearning_rateinput_pixeltrain_timetrain_accval_acc
9591CONV6431samereluXavier0.0001None---0.233050Adam5e-0510020886.1290.23
9692CONV6432samereluXavier0.0001Max31valid0.2---------
9793CONV12831samereluXavier0.0001None---0.3---------
9894CONV12832samereluXavier0.0001Max31valid0.3---------
9995CONV25631samereluXavier0.0001None---0.3---------
10096CONV25632samereluXavier0.0001Max31valid0.4---------
10197FC128---relu-0.0001----0.2---------
10298FC64---relu-0.0001----0.2---------
\n", "
" ], "text/plain": [ " model layer type size ... input_pixel train_time train_acc val_acc\n", "95 9 1 CONV 64 ... 100 208 86.12 90.23\n", "96 9 2 CONV 64 ... - - - -\n", "97 9 3 CONV 128 ... - - - -\n", "98 9 4 CONV 128 ... - - - -\n", "99 9 5 CONV 256 ... - - - -\n", "100 9 6 CONV 256 ... - - - -\n", "101 9 7 FC 128 ... - - - -\n", "102 9 8 FC 64 ... - - - -\n", "\n", "[8 rows x 24 columns]" ] }, "metadata": { "tags": [] }, "output_type": "display_data" } ], "source": [ "# Loading the Excel sheet with pre-defined hyperparameters:\n", "params_sheet = pd.read_excel(root_dir + '/Model Params.xlsx').drop(['Unnamed: 0'],axis=1)\n", "\n", "# Optional preview of the parameters for a selected model:\n", "model_number = 9\n", "display(params_sheet[params_sheet['model'] == model_number])" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "9WjsTOA9EAo1" }, "source": [ "### 1.2. Helper Functions\n", "\n", "We can now start building the helper functions that will load the data, compile the model and visualize the performance:\n", "\n", "#### 1.2.1. Model Compiler\n", "\n", "Given a model number (from available models on the Excel sheet), this function reads the hyperparameters from the Excel sheet and compile a sequantial model in Keras:" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "mLp6MFwbYWDa" }, "outputs": [], "source": [ "# A function to construct the network based on parameters saved on the excel sheet:\n", "def model_compiler(params_sheet, model_number):\n", " \n", " # Filtering the params_sheet to parameters specific to one given model number\n", " param_df = params_sheet[params_sheet['model'] == model_number].reset_index()\n", " \n", " # Loding hyperparams\n", " num_epochs = int(param_df.loc[0, 'num_epochs'])\n", " mini_batch_size = int(param_df.loc[0, 'batch_size'])\n", " num_classes = int(param_df.loc[0, 'num_classes'])\n", " optimizer_type = param_df.loc[0, 'optimizer']\n", " learning_rate = param_df.loc[0, 'learning_rate']\n", " input_pixel = param_df.loc[0, 'input_pixel']\n", " \n", " # Starting to construct the sequential model\n", " model = Sequential()\n", " \n", " # The CONV layers\n", " for i in param_df[param_df['type'] == 'CONV'].index:\n", " \n", " if param_df.loc[i, 'kernel_init'] == 'Xavier':\n", " initializer = keras.initializers.glorot_normal(seed = 1)\n", " else:\n", " print('Initializer Undefined')\n", " break\n", " \n", " # For layer 1 we need to define the input_shape\n", " if i == 0: \n", " model.add(Conv2D(filters = param_df.loc[i, 'size'], \n", " kernel_size = param_df.loc[i, 'filter_size'], \n", " strides = (param_df.loc[i, 'filter_s'], param_df.loc[i, 'filter_s']), \n", " padding = param_df.loc[i, 'filter_p'], \n", " kernel_initializer = initializer, \n", " kernel_regularizer = regularizers.l2(param_df.loc[i, 'L2_reg']),\n", " input_shape = (input_pixel, input_pixel, 1)))\n", " else:\n", " model.add(Conv2D(filters = param_df.loc[i, 'size'], \n", " kernel_size = param_df.loc[i, 'filter_size'], \n", " strides = (param_df.loc[i, 'filter_s'], param_df.loc[i, 'filter_s']), \n", " padding = param_df.loc[i, 'filter_p'], \n", " kernel_initializer = initializer, \n", " kernel_regularizer = regularizers.l2(param_df.loc[i, 'L2_reg'])))\n", " \n", " # Using batch normalization\n", " model.add(BatchNormalization())\n", " \n", " model.add(Activation(param_df.loc[i, 'activ_fun']))\n", " \n", " # Add pooling\n", " if param_df.loc[i, 'pool_type'] == 'Max':\n", " model.add(MaxPooling2D(pool_size = (param_df.loc[i, 'pool_size'], param_df.loc[i, 'pool_size']), \n", " strides = (param_df.loc[i, 'pool_s'], param_df.loc[i, 'pool_s']), \n", " padding = param_df.loc[i, 'pool_p']))\n", " if param_df.loc[i, 'pool_type'] == 'Average':\n", " model.add(AveragePooling2D(pool_size = (param_df.loc[i, 'pool_size'], param_df.loc[i, 'pool_size']), \n", " strides = (param_df.loc[i, 'pool_s'], param_df.loc[i, 'pool_s']), \n", " padding = param_df.loc[i, 'pool_p']))\n", " model.add(SpatialDropout2D(rate = param_df.loc[i, 'dropout_rate']))\n", " \n", " # Flattening\n", " model.add(Flatten())\n", " \n", " # The fully connected layers\n", " for i in param_df[param_df['type'] == 'FC'].index:\n", " model.add(Dense(units = param_df.loc[i, 'size'], \n", " activation = param_df.loc[i, 'activ_fun']))\n", "\n", " model.add(Dropout(rate = param_df.loc[i, 'dropout_rate']))\n", " \n", " # SoftMax layer as the output\n", " model.add(Dense(units = num_classes, activation='softmax'))\n", "\n", " # Compiling the model\n", " if optimizer_type == 'Adam':\n", " model.compile(optimizer = keras.optimizers.Adam(lr = learning_rate, epsilon = 1e-8),\n", " loss = 'categorical_crossentropy', \n", " metrics = ['accuracy'])\n", " return model, num_epochs, mini_batch_size, input_pixel" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "7TNGPg48EApZ" }, "source": [ "#### 1.2.2. Performance Visualizer\n", "Given the `history` output from the trained model, this function prints out the training, validation and test accuracies, as well as the accuracy plots and confusion matrix for the trained model:" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "Sc3LVK0eYWDh" }, "outputs": [], "source": [ "# A function to visualize the accuracies, confusion matrix and classification report:\n", "def summary_visualizer(history, path, X_test, y_test, params_sheet, model_number): \n", " \n", " param_df = params_sheet[params_sheet['model'] == model_number].reset_index()\n", " num_classes = int(param_df.loc[0, 'num_classes'])\n", " \n", " # Selecting the right expression names based on num of classes\n", " if num_classes == 3:\n", " expression_names = ['Happy', 'Sad', 'Surprised']\n", " if num_classes == 5:\n", " expression_names = ['Neutral','Happy', 'Sad', 'Surprised','Angry']\n", " \n", " clear_output(wait = True)\n", " \n", " # Defining true and predicted labels\n", " y_true = np.argmax(y_test, axis = 1)\n", " y_pred = np.argmax(model.predict(X_test), axis = 1)\n", " \n", " print(f\"Training accuracy:\\t({(history.history['acc'][-1]*100):.2f}%, {(max(history.history['acc'])*100):.2f}%) (last epoch, highest)\")\n", " print(f\"Validation accuracy:\\t({(history.history['val_acc'][-1]*100):.2f}%, {(max(history.history['val_acc'])*100):.2f}%) (last epoch, highest)\")\n", " print(f\"Test accuracy:\\t\\t({(sum(y_true == y_pred) / len(y_true) * 100):.2f}%)\\n\")\n", "\n", " # Constructing the confusion matrix and calculating percentage of correct and incorrect predictions\n", " conf_matrix = confusion_matrix(y_true = y_true, y_pred = y_pred)\n", " conf_matrix_df = pd.DataFrame(data = conf_matrix, columns = expression_names, index = expression_names)\n", " conf_matrix_percent = round(conf_matrix_df/conf_matrix_df.sum(axis = 1)*100, 2)\n", " \n", " print('Model accuracy and loss function:\\n')\n", "\n", " plt.figure(figsize = (14,4))\n", " gridspec.GridSpec(1,2)\n", "\n", " # Summarize history for accuracy\n", " plt.subplot2grid((1,2), (0,0))\n", " plt.plot(history.history['acc'])\n", " plt.plot(history.history['val_acc'])\n", " plt.title('Model Accuracy')\n", " plt.ylabel('Accuracy')\n", " plt.xlabel('Epoch')\n", " plt.legend(['Train', 'Test'], loc='lower right')\n", "\n", " # Summarize history for loss\n", " plt.subplot2grid((1,2), (0,1))\n", " plt.plot(history.history['loss'])\n", " plt.plot(history.history['val_loss'])\n", " plt.title('Model Loss')\n", " plt.ylabel('Loss')\n", " plt.xlabel('Epoch')\n", " plt.legend(['Train', 'Test'], loc='upper right');\n", " plt.savefig(path + os.sep + \"Accuracy_Plot.png\")\n", " plt.show();\n", "\n", " print('The confusion matrix:\\n')\n", "\n", " plt.figure(figsize = (12,4))\n", " gridspec.GridSpec(1,2)\n", "\n", " # Plotting the confusion matrix\n", " plt.subplot2grid((1,2), (0,0))\n", " sns.heatmap(conf_matrix_df, annot=True, cmap = 'Blues', fmt='g', linewidths = .8)\n", " plt.ylabel('True label', labelpad = 5)\n", " plt.xlabel('Predicted label', labelpad = 20)\n", " plt.title('Confusion matrix')\n", " plt.yticks(rotation=0);\n", "\n", " # Plotting the Percentage of correct and incorrect predictions for each class\n", " plt.subplot2grid((1,2), (0,1))\n", " sns.heatmap(conf_matrix_percent, annot=True, cmap = 'Blues', fmt='g', linewidths = .8)\n", " plt.ylabel('True label', labelpad = 5)\n", " plt.xlabel('Predicted label', labelpad = 20)\n", " plt.title('Percentage of correct/incorrect predictions for each class\\n(Predicted / Total in True Class)', fontsize = 10)\n", " plt.yticks(rotation=0);\n", "\n", " plt.tight_layout(w_pad=8)\n", " plt.savefig(path + os.sep + \"Confusion_matrix.png\")\n", " plt.show()\n", " \n", " # Printing the classification report\n", " print('\\nThe classification report:\\n')\n", " print(classification_report(y_true = y_true, \n", " y_pred = y_pred, \n", " target_names = expression_names))" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "EFTuOWYrEApb" }, "source": [ "#### 1.2.3. Data Loader\n", "This function simply loads the `.npy` files that we exported in the previous notebook, based on the number of classes that is defined on the Excel sheet (for a given model number):" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "v9Tk8dS0YM79" }, "outputs": [], "source": [ "# A function to load the appropriate data files:\n", "def data_loader(params_sheet, model_number):\n", " param_df = params_sheet[params_sheet['model'] == model_number].reset_index()\n", " num_classes = int(param_df.loc[0, 'num_classes'])\n", " \n", " if num_classes == 3:\n", " X_train = np.load(root_dir + '/Data/3 Expressions/X_train_3classes.npy')\n", " X_test = np.load(root_dir + '/Data/3 Expressions/X_test_3classes.npy')\n", " X_val = np.load(root_dir + '/Data/3 Expressions/X_val_3classes.npy')\n", " y_train = np.load(root_dir + '/Data/3 Expressions/y_train_3classes.npy')\n", " y_test = np.load(root_dir + '/Data/3 Expressions/y_test_3classes.npy')\n", " y_val = np.load(root_dir + '/Data/3 Expressions/y_val_3classes.npy')\n", "\n", " elif num_classes == 5:\n", " X_train = np.load(root_dir + '/Data/5 Expressions/X_train_5classes.npy')\n", " X_test = np.load(root_dir + '/Data/5 Expressions/X_test_5classes.npy')\n", " X_val = np.load(root_dir + '/Data/5 Expressions/X_val_5classes.npy')\n", " y_train = np.load(root_dir + '/Data/5 Expressions/y_train_5classes.npy')\n", " y_test = np.load(root_dir + '/Data/5 Expressions/y_test_5classes.npy')\n", " y_val = np.load(root_dir + '/Data/5 Expressions/y_val_5classes.npy')\n", " \n", " print(f'X_train shape:\\t{X_train.shape}')\n", " print(f'y_train shape:\\t{y_train.shape}')\n", " print(f'X_val shape:\\t{X_val.shape}')\n", " print(f'y_val shape:\\t{y_val.shape}')\n", " print(f'X_test shape:\\t{X_test.shape}')\n", " print(f'y_test shape:\\t{y_test.shape}')\n", "\n", " return X_train, X_test, X_val, y_train, y_test, y_val" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "9xCoUf_TEApo" }, "source": [ "### 1.3. Model Training\n", "\n", "With the helper functions defined, we can now train the CNNs, first with 3 expressions (Happy, Sad, Surprised) and then with 5 expressions (Happy, Sad, Surprised, Angry and Neutral):\n", "\n", "#### 1.3.1. Model with 3-Classes (Happy, Sad, Surprised)\n", "\n", "For 3 classe, training is done on different combinations of hyperparameters and the best performing model is found to be Model # 9 on the Excel sheet with the following key parameters and performance metrics:\n", "\n", "\n", "\n" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "SOhzKqfBEApo" }, "outputs": [], "source": [ "# Loading the data (3 classes):\n", "model_number = 9 # See the excel sheet for list of other models\n", "X_train, X_test, X_val, y_train, y_test, y_val = data_loader(params_sheet, model_number)" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "3RLc3osmEApp" }, "outputs": [], "source": [ "# Compiling the model (3 classes):\n", "model, num_epochs, mini_batch_size, input_pixel = model_compiler(params_sheet, model_number)" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 932 }, "colab_type": "code", "executionInfo": { "elapsed": 1343, "status": "ok", "timestamp": 1560214368638, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "t321xtBQtNyF", "outputId": "53625a7c-ad8b-4def-beb4-b4bde59c1c13", "scrolled": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Training accuracy:\t(86.12%, 86.12%) (last epoch, highest)\n", "Validation accuracy:\t(90.15%, 90.23%) (last epoch, highest)\n", "Test accuracy:\t\t(90.19%)\n", "\n", "Model accuracy and loss function:\n", "\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0MAAAEWCAYAAACkHEyCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xd8nOWV9//PmVHvVnGR5Y5t3G0w\nkFBC7y0k9IcUEsJmfyFlCdmQPFlCsskTUnY3WSCbkAQCyQYCJATTYmroxQZs4wa4YcuSLVm2ZElW\nn/P74x7ZgyzbsjWjUfm+X6/7NXefIxl0z5nrus5l7o6IiIiIiMhQE0p2ACIiIiIiIsmgZEhERERE\nRIYkJUMiIiIiIjIkKRkSEREREZEhScmQiIiIiIgMSUqGRERERERkSFIyJBLDzMabmZtZSg/O/ayZ\nvdQXcYmIiPSWnnEie1MyJAOWmW0ws1YzK+6y/+3oH/vxyYnsQ7HkmFmDmT2R7FhERGTg6M/PuINJ\nqkT6OyVDMtCtB67o3DCzWUBW8sLZyyeBFuB0MxvZl2+sh5SIyIDX359xIgOekiEZ6P4AfDpm+zPA\nPbEnmFm+md1jZtVm9oGZfcfMQtFjYTP7mZltM7N1wLndXPs7M6s0s81m9gMzCx9EfJ8BfgUsA67q\ncu8xZvbXaFw1ZnZbzLEvmNkqM6s3s5VmdkR0v5vZYTHn/d7MfhBdP8nMys3sm2a2BbjLzIaZ2aPR\n99gRXS+Lub7QzO4ys4ro8b9F9y83s/NjzkuN/o7mHcTPLiIivdPfn3F7MbN0M/t59LlSEV1Pjx4r\njj6Has1su5m9GBPrN6Mx1JvZu2Z2am/iEOkpJUMy0L0G5JnZtOgf8MuBP3Y551YgH5gInEjwYLk6\neuwLwHnAPGA+cHGXa38PtAOHRc85A7imJ4GZ2TjgJOB/o8unY46FgUeBD4DxwGjgvuixS4Cbo+fn\nARcANT15T2AkUAiMA64l+H/8ruj2WKAJuC3m/D8QfMs4AxgO/Fd0/z18OHk7B6h097d7GIeIiPRe\nv33G7cf/BT4CzAXmAEcD34ke+zpQDpQAI4BvA25mU4HrgKPcPRc4E9jQyzhEekTJkAwGnd+cnQ6s\nAjZ3Hoh5eHzL3evdfQPwH8CnoqdcCvzc3Te5+3bgRzHXjiBIAr7m7o3uXkWQLFzew7g+BSxz95UE\nic6MmJaVo4FS4BvReze7e+dA1WuAn7j7Ig+scfcPevieEeC77t7i7k3uXuPuf3H3Xe5eD/yQ4GGJ\nmY0Czga+6O473L3N3Z+P3uePwDlmlhfzs/yhhzGIiEj89Ndn3L78H+D77l7l7tXA92LiaQNGAeOi\nz5wX3d2BDiAdmG5mqe6+wd3X9jIOkR7RmAIZDP4AvABMoEv3AaAYSCVogen0AUFLDAQJyaYuxzqN\ni15baWad+0Jdzt+fTwO/AXD3zWb2PEEXh7eBMcAH7t7ezXVjgEN9CFS7e3PnhpllETzczgKGRXfn\nRh+gY4Dt7r6j603cvcLMXgY+aWYPESRNXz3EmERE5ND112fcvpR2E09pdP2nBD0fnoy+5x3ufou7\nrzGzr0WPzTCzhcD17l7Ry1hEDkgtQzLgRVtN1hN8w/XXLoe3EXwTNS5m31j2fLNWSZAUxB7rtImg\n+EGxuxdElzx3n3GgmMzsWGAy8C0z2xIdw3MMcGW0sMEmYOw+ihxsAibt49a7+PDg2a5FGbzL9teB\nqcAx7p4HfKwzxOj7FJpZwT7e626CrnKXAK+6++Z9nCciIgnSH59xB1DRTTwV0Z+l3t2/7u4TCbqA\nX985Nsjd/+Tux0evdeDHvYxDpEeUDMlg8XngFHdvjN3p7h3A/cAPzSw3Oo7nevb0ub4f+IqZlZnZ\nMODGmGsrgSeB/zCzPDMLmdkkMzuxB/F8BngKmE7Qb3ouMBPIJGhleYPgIXWLmWWbWYaZHRe99rfA\nDWZ2pAUOi8YNsIQgoQqb2VlEu7ztRy7BOKFaMysEvtvl53sC+GW00EKqmX0s5tq/AUcQtAh1/TZS\nRET6Tn97xnVKjz6/OpcQcC/wHTMrsaAs+E2d8ZjZedFnmgF1BN3jImY21cxOiRZaaCZ4bkUO8nck\nckiUDMmg4O5r3X3xPg5/GWgE1gEvAX8C7owe+w2wEFgKvMXe37p9GkgDVgI7gAcJ+jvvk5llEPTT\nvtXdt8Qs6wm6O3wm+gA7n2DQ6kaCAaWXRX+WBwjG9vwJqCdISgqjt/9q9Lpagn7Zf9tfLMDPCRKw\nbQQDcf/e5finCL5VXA1UAV/rPODuTcBfCLpmdP29iIhIH+lPz7guGggSl87lFOAHwGKCKqrvRN/3\nB9HzJwNPR697Ffiluz9HMF7oFoJn1RaCgj7fOog4RA6ZBePWRET2ZmY3AVPc/aoDniwiIiIywKiA\ngoh0K9qt7vPsqQIkIiIiMqiom5yI7MXMvkAwuPYJd38h2fGIiIiIJIK6yYmIiIiIyJCkliERERER\nERmSBtyYoeLiYh8/fnyywxARGdLefPPNbe5ekuw4+iM9p0REkq+nz6kBlwyNHz+exYv3VV1SRET6\ngpl9cOCzhiY9p0REkq+nzyl1kxMRERERkSFJyZCIiAxJZnanmVWZ2fJ9HD/czF41sxYzu6Gv4xMR\nkcRTMiQiIkPV74Gz9nN8O/AV4Gd9Eo2IiPS5ATdmSEREJB7c/QUzG7+f41VAlZmd22dBiYj0Ultb\nG+Xl5TQ3Nyc7lD6RkZFBWVkZqamph3S9kiEREZFeMrNrgWsBxo4dm+RoRGQoKy8vJzc3l/Hjx2Nm\nyQ4nodydmpoaysvLmTBhwiHdQ93kREREesnd73D3+e4+v6REFcdFJHmam5spKioa9IkQgJlRVFTU\nq1awhCZDZnaWmb1rZmvM7MZujo8zs2fMbJmZ/cPMyhIZj4iIiIjIYDcUEqFOvf1ZE9ZNzszCwO3A\n6UA5sMjMFrj7ypjTfgbc4+53m9kpwI+ATyUqJhGRXnGHph3QuA0aq6ChChqroakWUtIhLRvSciAt\nK2Y9+8PrKZkQSmKjfKQD6rdAezMUTUpeHNKttzbuYOHyLXzzrMMJhYbOhxkRkWRJ5Jiho4E17r4O\nwMzuAy4EYpOh6cD10fXngL8lMB6R/ssdmmuhuS5m2dlluw5aovta6oMPtd4BHtmzHulm2wzySqFg\nLOSPDV4LxkLBGMgthfB+/gy4B++3czPUlUPdpujrZthZAamZkF0COSWQPRxyhkN28Z71rCIIhbu/\nZ9MOaNoOu6KvTTtg1/ZgPZwGBeOCGAvGQv4YyMiLz++6vQVaGqA1urQ0QGv9nn27X3dCQ3WQ9DRW\nR9erIdLW+xhSs4IlLQtSs6Ov+9iXngPp+cHPn5635zU9FzLyg/XOf8PO/446/406/712//tthvoK\niLTDmGPg80/2/mcZwMzsXuAkoNjMyoHvAqkA7v4rMxsJLAbygIiZfQ2Y7u47ExXT6sp6fv3COj71\n0XGUDctK1NuIiCRMTU0Np556KgBbtmwhHA7T2X34jTfeIC0t7YD3uPrqq7nxxhuZOnVqQmOFxCZD\no4FNMdvlwDFdzlkKfAL4BXARkGtmRe5eE3uSBqbKoNTWDBtegvf+Du8thLqN+z8/vfPDcH7QwhBK\nCRYLBQmHhWNeQ3u2Ix1B4vL+09Cw5cP3tDDkj96TJOWODJKR3R+my4NEIVYoJUiuckuDRKFqZdBC\n0m2SYEFClDM8+AC+K5r0eMd+fs78oNWio+XD+zMKYhK5sXuSpPTcIAFoqt2TUHau77WvrufJTDgt\nSOqyiyFnBIyYFU36OhO/mPXMYdDRCq2N0SSrMVjaGvesx+5vbYS2XdDWtGe9dVfw79O6K7od3d/R\neuBYU7OC/zY6E7xYodTg3yt/DIw7NvrvXQZFk3v2exjE3P2KAxzfAvRp9+2JJdkArN/WqGRIRAak\noqIilixZAsDNN99MTk4ON9zw4ana3B13J7SPnhJ33XVXwuPslOxqcjcAt5nZZ4EXgM3AXp+S3P0O\n4A6A+fPne18GKBJX9Vvg/SeD5Gftc8GH5ZRMmHQyHP2FIHHIyI8ueXvW0/P2bmE5FG3NQStB7QdQ\nuxFqN0VfN8L656G+EjILox+WJ8HEE4P1vNHBh+n8siCx6ba1pzboPtZQFW1N6bIeCgf3zioMXjOH\n7b2eURC0ckQiQStM3aa9Y61ZA2ufDRKF7lg4+J1lFgT3y8gPEqfO9fQcSMuNvubsed29Hj2Wkn5w\nv9twStCaQ5wHz7e3BC2Bu1sGd+7jtS74GfLLoglP9N8re3hyu+XJQZlYHCRD66obOWGyCjGIyOCx\nZs0aLrjgAubNm8fbb7/NU089xfe+9z3eeustmpqauOyyy7jpppsAOP7447ntttuYOXMmxcXFfPGL\nX+SJJ54gKyuLhx9+mOHDh8ctrkQmQ5uBMTHbZdF9u7l7BUHLEGaWA3zS3WsTGJPIodu+DhbfGXSX\nyi6OLtHWgayY7bSYb3MjEdiyNEh+3vs7VLwd7M8rg7lXwJSzYPzxQXezvpCaESQ5+xor4h50qztY\nZkFCkzkMiuPQ4hAKQe6IYCmb332cu2qC5Ki1MZr45AcJT3ruof0M/VVKerBkFyc7EukDJbnpZKeF\nWb+tMdmhiMgg8L1HVrCyIr49e6eX5vHd82cc0rWrV6/mnnvuYf784Nl+yy23UFhYSHt7OyeffDIX\nX3wx06dP/9A1dXV1nHjiidxyyy1cf/313Hnnndx441512Q5ZIpOhRcBkM5tAkARdDlwZe4KZFQPb\n3T0CfAu4M4HxSKJteQfeuid4HT4dSudB6VwoORzChzYRFhAkFPUVQZetnOHBN969ud/BcIcPXoHX\nfgmrHwtaN3JLg1aL9qbur0nNhuyiIDGq2xztmmZQdhSc8m9BAjRiRv/8wN4fY+qO2Z6EVGQQMTMm\nluSwTsmQiAxCkyZN2p0IAdx777387ne/o729nYqKClauXLlXMpSZmcnZZ58NwJFHHsmLL74Y15gS\nlgy5e7uZXQcsBMLAne6+wsy+Dyx29wUEA1d/ZGZO0E3uS4mKRxKkuQ7eeRDe/kPQ6hFOh1Gz4Z0H\nYPHvgnNSMmDEzCAxKp0HozoTpJS977VtTdANquZ92PY+1KwNtmMTDwsHA+uHjYdhE4LXwgl71uMx\n0L69FVb8NUiCKpcGLR4nXA9HfQHyRgXntDYGSVFjTfQ1uuyKbjdUwdgxMOVMmHyGPriLSI9MKM7m\nrY07kh2GiAwCh9qCkyjZ2dm7199//31+8Ytf8MYbb1BQUMBVV13V7XxBsQUXwuEw7e3tcY0poWOG\n3P1x4PEu+26KWX8QeDCRMUgCuMPG14JWoBUPBYnK8Blw9k9g1iXB2I9IBHasDxKkirehYgks/TMs\n+m1wj5RMGDkTCicGY0Fq1gRjSzpZGIaNg6LDYMLHgm5d+WOCc7avhx0bgvuvfDgY8B8rqyhIjEoO\nD1pgRkwPkrGeJCONNfDmnfDGb4MWneKpcN7PYfZlH+7+BntKJg8b35vfpojIh0wozuaRZRU0t3WQ\nkRqHsYIiIv3Qzp07yc3NJS8vj8rKShYuXMhZZ53V53Eku4CCDCQN1bD03iAJqnk/GGg+5zI44tNQ\nesSHu1iFQnvGpsy6ONgXicD2tUFiVPE2VC4JqqkVjA1aT4onB8lP0eQgwUg5cOlFIGhR2rEhWLav\nD5Kk7euCQgVL/rjnvOzh0eQougyfHiRMqRlQtRpe/x9Yel9QyWzSKXDh7cGrBp+LSB+aWJIdfOe0\nfRdTRuQmOxwRkYQ44ogjmD59Oocffjjjxo3juOOOS0oc5j6wirPNnz/fFy9enOwwBr7OpKGjPSg1\n3NEWlD7uaOt+e/Nb8O7j0flJPhIkQDM+HrSM9GcN1VC1ArbGLNWrg4QHoqWly4KKZSkZQQvQR/4Z\nhk9Lbtwi/ZyZvenu3VS3kN4+p94pr+P8217iV1cdwVkzR8UxMhEZClatWsW0aUPrc0x3P3NPn1Nq\nGRoq3IOWmNWPBUvVygNfEyurCI75YpAElSR+Aqy4ySmBnJNg4kl79kU6gpajrcth60rY9i7M+xTM\nv1rjekQk6cYXB11yVURBRCTxlAz1V51VzJb8b1BWd/j0YBkxPRjM3xMdbfDBy9EE6HHYWR5M0Dn2\nWDjzR0FBg3BaMIlmODWYnDGcEn3tsp2SMXi6i4XCQZe84skw46JkRyMi8iG5GamU5KazvlrJkIhI\noikZ6m/ammH5X4LxK1vegfT8YH9LTNXx3FHR5GhadOzLtOjYl8ygwtmaZ4IE6L2/BxNhpmTApFPh\n5G8HZZ2zi5Lzs4mISI9MLM5Wy5CISB9QMtRf7KwMSlEvvgt2bYOSaXD+L2DWpUGSs7MCqlYF41+q\nVgVjX954CTpagustBAXjoL4yGA+TUQBTz4bDz4NJJ/f/sT0iIrLbxJJsFq7YmuwwREQGPSVDyVb+\nZtAKtOKhYCzLlLPgI1+ECSd+uDpb/uhgmXzann2dY1+qVgZjX6pXBfPZTDsv6ArXdR4fEREZECYU\nZ7O9sZXaXa0UZPWwsqaIiBw0fVpOho62YH6c138F5YsgPQ+OvhaO/kIw705PxY59mX5h4uIVEZE+\nNbE4B4D12xqZN1bJkIhIoigZ6kuRCCy7D579AezcDIWTgolK514J6ZpLQkREAhNKgq7N66obmTe2\nh0VzRET6gZqaGk499VQAtmzZQjgcpqSkBIA33niDtLSefcFz5513cs455zBy5MiExQpKhvrOxtfg\n7zcGk42OPhLO+zkcdtrgqdAmIiJxM2ZYFuGQsV5FFERkgCkqKmLJkiUA3HzzzeTk5HDDDTcc9H3u\nvPNOjjjiCCVDA17tRnj65qBCXG4pfOI3MPNiJUEiknCRiLOzuY0du9rYsSsYf7KjMVgvyErj4iPL\nkh2i7ENaSoixhVlKhkRkULn77ru5/fbbaW1t5dhjj+W2224jEolw9dVXs2TJEtyda6+9lhEjRrBk\nyRIuu+wyMjMzD6pF6WApGUqUlgZ4+efwyq2AwYk3wnFfUVU3kUGqM/GoaWxl+z6WmsZW6na1khoO\nkZkWJistTHZayu71rLSU6Oue9YhDS3sHLe0RWtqir+0RWtsjMfuD9frm9mjSEyQ8dU1tRLz7eOeU\n5SsZ6ucmFGeztroh2WGIyED2xI3BVC3xNHIWnH3LQV+2fPlyHnroIV555RVSUlK49tprue+++5g0\naRLbtm3jnXeCOGtraykoKODWW2/ltttuY+7cufGNvwslQ/EWicCyP8Mz3wvKXM+6BE67GfL1oUMk\nWVrbIzS1dtDc3kF7xOnocNojEToiHmzvfo3Q3hFst3REaGhup6GlnfrmNhqa29nZ3E59czsNLW3R\n12C7Ptr60rGPzCMrLUxhdhpF2WnkZ6XR3hGhvrmdqp0tNLa209Tawa7WDpraOnr8M6WEjLSUEOkp\nIdJTwqSlhMjNSGFYVhqlBZkMy0qjICuVgqw0hmWl7t4elpXGsKw0cjP059/M7gTOA6rcfWY3xw34\nBXAOsAv4rLu/1VfxTSjO5pW124hEnFDIDnyBiEg/9vTTT7No0SLmz58PQFNTE2PGjOHMM8/k3Xff\n5Stf+QrnnnsuZ5xxRp/GpadhPG18PTou6K1gXNCl98CYo5MdlUi/0tLewY7Gtj0tJrta2RFtNdkR\n3W5p68DMCBmEzAiZYbvXg1eirx0RZ1dre5BMxCQVsfva99U8chDMICc9hbyMVHLSU8jNSKEwO41x\nRdnBelYaw6IJT2GXJSM13KP3iEQ8Gvue+MMh253wpKeESE8NkRYOkRJWV9s4+D1wG3DPPo6fDUyO\nLscA/xN97RMTirNpbouwZWczpQWZffW2IjKYHEILTqK4O5/73Of493//972OLVu2jCeeeILbb7+d\nv/zlL9xxxx19FpeSoXhoroNHr4flD0LuKLjo18FkqRoXJAOMu9PcFtmTSMR8MO9MNDqP7Wrt7LLV\nEe2yFXTXau0IunMFr3u6cu1sbmNHYxsNLe3dvrcZFGSmMiw7jYyUMB6NJ+JOxCHijkdfO9fdg+uy\n0sJkpqWQlRpmVH5qt93OMtNSyEgNkRoKEQ4ZKWELXkNGOBSKvgb7U0IhUsJGbnoKuRmp5GSkkJ0W\nxiyx386HQkZ2egrZ6SlAekLfS8DdXzCz8fs55ULgHnd34DUzKzCzUe5e2RfxTYypKKdkSEQGutNO\nO42LL76Yr371qxQXF1NTU0NjYyOZmZlkZGRwySWXMHnyZK655hoAcnNzqa+vT3hcSoZ6q6MdHrga\n1j8PH/tXOP5rGhckfcrd2dXaQW1TG3W72qhtaqVuVxt1TW3UNrWxs6ntQ0lM8+4EJ9qS0vbhblp+\nkI0oe7pq7Wm9iO2+lZ2ewrCsEFNG5DIsK42inKCbVmF2KoXZ6RRmd3bhSiOsrkDSv4wGNsVsl0f3\n7ZUMmdm1wLUAY8eOjcub75lrqIHjJxfH5Z4iIskya9Ysvvvd73LaaacRiURITU3lV7/6FeFwmM9/\n/vO4O2bGj3/8YwCuvvpqrrnmGhVQ6PcWfhvWPgPn/zcc+ZlkRyODRHNbB9X1LVTVt7CtoYXq+ugS\nXa9paNmd6NTuattvN7CUkH2olSQjNWgpyc1IYUReOllpKbv3BS0oYbJSg/Mz08Jkp4fJTE358PHo\nvdJTQglvLREZCNz9DuAOgPnz5/e+XyZE//8Ms04V5URkgLr55ps/tH3llVdy5ZVX7nXe22+/vde+\nSy+9lEsvvTRRoe2mZKg3Fv0W3vg1fORLSoRkvzoiTu2uYFxMTUMrNY0tbG9sZVtDK9s71+tbdyc+\n9d10JTODwqw0SnLTKcxOY1p+JvlZqeRnplKQmUpBdD0/M233ekFWKpmpie/eJTJIbQbGxGyXRff1\nCTNjQnE266qVDImIJIqSoUO19jl4/F9h8hlwxt4DwWTw64g42xtbP9RiU92lJWdbQ5Do7NjV2m2J\n485xMkU56RRlpzG9NI+S3HSKc9IpyY0u0fXC7DRSNWhepC8tAK4zs/sICifU9dV4oU4TirNZVl7X\nl28pIjKkKBk6FNvehwc+A8VT4JO/g1DPKkVJ/xaJOLVNbdGWmuC1s8JZ7Nwx2xqCBGh7Y0u3CU5W\nWnh3EjOxJJujJhRSHK0q1pn0FOUEyc2wrFRVBRNJEjO7FzgJKDazcuC7QCqAu/8KeJygrPYagtLa\nV/d1jBOLs3n8nUpa2jtIT9GzRkR6pnP8zVDgBzvYuQslQwdr13b402UQSoEr74OMvGRHJAeprSPC\nu1vqWVpey7JNdbyzuY4tO5up3UfrDUBuegrDognN6IIM5o7J391iE9uKU5yTHq0EJiL9nbtfcYDj\nDnypj8Lp1sSSHCIOG2t2MXlEbjJDEZEBIiMjg5qaGoqKigZ9QuTu1NTUkJGRccj30Ke2g9HRFrQI\n1W2CTy+AYeOTHZEcQCTirNvWwNJNdSwrr2VpeR0rK3fS2h4BID8zldll+cwbW/CheWGKstMZlp26\n+1XfyIpIMkwojpbX3taoZEhEeqSsrIzy8nKqq6uTHUqfyMjIoKys7JCvVzLUU+7w+Ddg/Qvw8V/B\nuI8mOyLpwt3ZtL2JZZtreae8jqXltSzfvHP3vDZZaWFmjs7nMx8dx+yyAmaX5TO2MGvQf2siIgPX\nhOhcQ+tVUU5Eeig1NZUJEyYkO4wBQ8lQT73+a3jzLjj+X2DufntWSB9wdyrrmllWXsc7m2tZVl7H\nsvI66praAEgLh5g2KpeL5o1mdlk+c8YUMKkkR/PYiMiAkpeRSnFOOutVUU5EJCGUDPXE+0/Dwm/B\n4efBKTclO5ohqW5XG4s/2M7S8jreKa/lnc11bGtoBYJ5dKaOzOWcWSOZNTpo8ZkyIpe0FBUmEJGB\nb2JxNuu2NSQ7DBGRQUnJ0IFUrYYHr4bhM+CiX0NIH7D7Qu2uVl5fv53X123ntXU1rNqyE3cIGUwe\nnstJU4czuyyfWaPzmTYqj4xUjekRkcFpQnE2z6zemuwwREQGJSVD+9NYA/deBikZQeW49JxkRzRo\n7WiMJj/ra3ht3XZWR5Of9JQQR44bxtdOncIxEwuZNTpf1dpEZEiZUJLNtsWt1DW1kZ+ZmuxwREQG\nFX2q3Jf2VvjzVbCzEq5+HPIPvUqF7K0j4ry+roYnV27ltXU1rN5SD0BGapD8XH/aFI6ZWMScMfmq\n5CYiQ9rE4j1FFOaOKUhyNCIig4uSoX15+x7Y+EowqWrZ/GRHMyh0RJzFG7bz6LJKnlheybaGVjJS\nQ8wfV8gNZ4ziIxOLmF1WoLE+IiIxJu6uKNegZEhEJM6UDO3LuuchfyzMujjZkQxokYjz1sYdPLqs\nksffqaSqvoWM1BCnHj6Cc2eP4uSpw8lMU8uPiMi+jCnMImSoopyISAIoGeqOO3zwMkw+M9mRDEju\nzpJNtTy2rJLH3qmksq6ZtJQQJ08t4bzZpZxy+HCN+xER6aH0lDBjCrNYq7mGRETiTp9Iu1O9GnbV\nwPjjkh1Jv+fuVNQ1896Wet7dWs97W+p5Y8N2ync0kRYO8bEpxXzzrMM5ddpwcjM08FdE5FBMKM5W\ny5CISAIoGerOhpeC1/HHJzeOfqamoWV3wvPu1gbei67Xt7TvPmdkXgYzR+fxtdOmcPr0Eap8JCIS\nBxOKs3l93XbcHTNNHi0iEi9Khrqz4SXIK4OCccmOJKmq61t47t0qnl1VxeIPdrCtoWX3sYKsVKaO\nyOWiI0YzZUQuU0fmMmV4LvlZSn5EROJtYkkOTW0dbNnZzKj8zGSHIyIyaCgZ6qpzvNCkU2GIffvm\n7qys3Mmzq6p4ZnUVS8trcYdR+RmcNLWEw0cGSc/UEbmU5Kbr20kRkT6yu7x2daOSIRGROEpoMmRm\nZwG/AMLAb939li7HxwJ3AwXRc25098cTGdMBbXsPGquHzHih5rYOXl1bw9OrtvLs6ioq65oxgzll\nBVx/2hROmTac6aPylPiIiCTRhGgytG5bI8ceVpzkaEREBo+EJUNmFgZuB04HyoFFZrbA3VfGnPYd\n4H53/x8zmw48DoxPVEw9suHF4HUQjxeqqm/m2VVVPL1qKy+t2UZzW4SstDAnTC7mX06fwslTh1OS\nm57sMEVEJGpkXgaZqWHWq6KCvg53AAAgAElEQVSciEhcJbJl6GhgjbuvAzCz+4ALgdhkyIG86Ho+\nUJHAeHpmw8uQWwrDJiQ7krhxd9ZUNfDkyq08vWorSzYF3d9GF2Ry2fwxnDJtBB+ZWEh6iub7ERHp\nj0IhY3xxNuuqG5IdiojIoJLIZGg0sClmuxw4pss5NwNPmtmXgWzgtO5uZGbXAtcCjB07Nu6B7uYe\nFE+YeNKAHy/U3hHhzQ928NTKrTy1aisf1OwCYHZZPtefNoXTpo/g8JG56v4mIkNaD7pzjwPuBEqA\n7cBV7l7e54ESjBtaUVGXjLcWERm0kl1A4Qrg9+7+H2b2UeAPZjbT3SOxJ7n7HcAdAPPnz/eERVOz\nBhqrBux4ocaWdl54r5qnVm3ludVV7NjVRlo4xEcnFfGFEyZy2rQRjMzPSHaYIiL9Qg+7c/8MuMfd\n7zazU4AfAZ/q+2iDcUN/X7GF1vYIaSmhZIQgIjLoJDIZ2gyMidkui+6L9XngLAB3f9XMMoBioCqB\nce3b7vFCJyTl7Q+Fu7O0vI773tjIgqUV7GrtID8zlVMOH87p00fwsSkl5KQnO+cVEemXetKdezpw\nfXT9OeBvfRphjIkl2XREnI3bd3HY8JxkhSEiMqgk8lPyImCymU0gSIIuB67scs5G4FTg92Y2DcgA\nqhMY0/5teBlyRkLhxKSF0FN1TW08vGQz976xiVWVO8lMDXP+nFFcNK+Mo8YPIyWsbw1FRA6gJ925\nlwKfIOhKdxGQa2ZF7l4Te1JfdOfurCi3flujkiERkThJWDLk7u1mdh2wkKAv9p3uvsLMvg8sdvcF\nwNeB35jZvxAUU/isuyeuG9z+Aw7GC40/vt+OF3J33tq4gz+9vonH3qmguS3CjNI8fvDxmVw4t5Tc\nDE14KiISZzcAt5nZZ4EXCL7c6+h6Ul90555YHCRA67c1ACMS8RYiIkNOQvtPRecMerzLvpti1lcC\n/WOAzvZ10LClX44Xqt3Vyl/f2sy9b2zk/aoGstPCfOKIMq44aiyzyvKTHZ6IyEB1wO7c7l5B0DKE\nmeUAn3T32j6LMEZ+VipF2Wmsq1Z5bRGReNFgkk79cLzQu1vq+fXza3n0nUpa2yPMGVPAjz85i/Nm\nl5KtcUAiIr11wO7cZlYMbI8W9vkWQWW5pJlQnM06zTUkIhI3+kTdacPLkD0cig5LdiQsK6/ltmfX\n8OTKrWSlhbls/hiuOHos00vzDnyxiIj0SA+7c58E/MjMnKCb3JeSFjBBMvSP95I3tFZEZLBRMgT9\nZrzQ6+tquO25Nbz4/jbyMlL4yqmTufrY8QzLTktaTCIig1kPunM/CDzY13Hty8SSHB54s5z65jaN\nExURiQMlQwA71kN9RVLGC7k7z79Xze3PrWHRhh0U56TxzbMO56qPjNWDTkREPiS2otzssoIkRyMi\nMvApGYKgVQj6dLxQJOI8uXILtz+3lnc21zEqP4Obz5/OZUeNJTMt3GdxiIjIwDGxRMmQiEg8KRmC\n6HihEiiekvC3cnceWVbJrc+8z/tVDYwvyuLHn5zFRfPKNKO4iIjs17iiLMxQRTkRkThRMtQ5Xmjc\ncX0yXuiX/1jLTxe+y9QRufzi8rmcO2uUJkgVEZEeSU8JUzYsUxXlRETiRMlQ7QewsxzGfy3hb/Xn\nRRv56cJ3+fjcUv7z0rmEQv1zclcREem/JhTnRCdeFRGR3lKTxO7xQscn9G2eXLGFb/31HT42pYSf\nXDxHiZCIiBySicXZrK9uxN2THYqIyICnZGjDy5BVBCWHJ+wt3li/nS/f+zazygr4n/9zhMYGiYjI\nIZtYkk1jawdV9S3JDkVEZMDTp/IEjxdavWUnn797EaOHZXLXZ48iO109E0VE5NB1ltdWEQURkd4b\n2snQjg+gbmPCusht2r6LT//uDbLTUrjnc0dTqMlTRUSkl2LnGhIRkd4Z2snQBy8HrwlIhmoaWvjM\nnW/Q3NbB3Z87mrJhWXF/DxERGXpK8zNJTwmxrlpFFEREemto99na8DJkDoOSaXG9bWNLO5/7/SI2\n1zbxx2uOYerI3LjeX0REhq5QyJhQnK2WIRGROBjaLUMbXgzGC4Xi92tobY/wxT++yfKKndx+5REc\nNb4wbvcWEREBlAyJiMTJ0E2GajcFcwzFsYtcJOLc8MBSXnx/Gz/6xCxOmz4ibvcWERHpNLEkm43b\nd9HWEUl2KCIiA9rQTYbiPF7I3fn3x1ayYGkF/3rWVC6dPyYu9xUREelqQnEO7RFn0/ZdyQ5FRGRA\nG7rJ0IaXIKMAhs+Iy+1++Y+13PXyBj533AT++cRJcbmniIhId1RRTkQkPoZ2MhSn8UKLNmznpwvf\n5cK5pXzn3GlYguYsEhERAZhUormGRETiYWgmQ3WbYcd6GH9cXG734OJystPC3PKJ2YRCSoRERAYK\nMzvLzN41szVmdmM3x8ea2XNm9raZLTOzc5IRZ1cFWWkMy0plnVqGRER6ZWgmQ3EcL9TS3sETyys5\nY8ZIMtPCvb6fiIj0DTMLA7cDZwPTgSvMbHqX074D3O/u84DLgV/2bZT7FlSU01xDIiK9MTSToQ0v\nQXo+jJjZ61u98N42dja3c8Gc0jgEJiIifehoYI27r3P3VuA+4MIu5ziQF13PByr6ML79mlCcozFD\nIiK9NHSToXHHQqj3LTkLllYwLCuV4ycXxyEwERHpQ6OBTTHb5dF9sW4GrjKzcuBx4Mvd3cjMrjWz\nxWa2uLq6OhGx7mViSTZbd7bQ0NLeJ+8nIjIYDb1kaGclbF8bl/FCjS3tPLVyC+fMGkVqeOj9KkVE\nhoArgN+7exlwDvAHM9vrD7673+Hu8919fklJSZ8E1llEYVXlzj55PxGRwWjofYKP43ihp1dtpbkt\noi5yIiID02YgdlK4sui+WJ8H7gdw91eBDKBfdAU49rBiMlJDPPR215BFRKSnhl4ytOElSM+DkbN7\nfasFSyoYlZ/BUeML4xCYiIj0sUXAZDObYGZpBAUSFnQ5ZyNwKoCZTSNIhvqmH9wB5GWkcs7MUTyy\npIKm1o5khyMiMiANzWRo7Ed7PV5oR2Mrz79XzflzSlVOW0RkAHL3duA6YCGwiqBq3Aoz+76ZXRA9\n7evAF8xsKXAv8Fl39+REvLdL5o+hvqWdv6+oTHYoIiIDUkqyA+hT9Vuh5n044lO9vtUTy7fQHnF1\nkRMRGcDc/XGCwgix+26KWV8JxGdSugQ4ZkIhYwuzuH9RORfNK0t2OCIiA84BW4bM7MtmNqwvgkm4\nOI4XWrB0MxOLs5lRmnfgk0VERBIgFDIuObKMV9fVsLFmV7LDEREZcHrSTW4EsMjM7o/O1D1w+4R9\n8Aqk5cLIOb26zZa6Zl5fv53z55QykH8dIiIy8H3yyDLM4ME3Nx34ZBER+ZADJkPu/h1gMvA74LPA\n+2b2/8xsUoJji78zfwjXPAXh3vUOfHRZBe5wwVx1kRMRkeQqLcjkhMklPPhmOR2RfjOcSURkQOhR\nAYXoYNEt0aUdGAY8aGY/SWBs8ZeSDsOn9fo2C5ZWMHN0HpNKcuIQlIiISO9cOr+MirpmXl6zLdmh\niIgMKD0ZM/RVM3sT+AnwMjDL3f8ZOBL4ZILj63fWb2tkWXmdCieIiEi/cfr0ERRkpfLnxeoqJyJy\nMHrSMlQIfMLdz3T3B9y9DcDdI8B5CY2uH3pkaQUA581WMiQiInHWtAPWv3DQl6WnhPn43NE8tWIr\nOxpbExCYiMjg1JNk6Alge+eGmeWZ2TEA7r4qUYH1R+7Ow0s2c/T4QkoLMpMdjoiIDDZv/xHuPh8a\nDn5e10vnj6G1I8LDSzYnIDARkcGpJ8nQ/wANMdsN0X0HFK0+966ZrTGzG7s5/l9mtiS6vGdmtT0L\nOzlWVu5kbXWjCieIiEhilM4LXiuXHPSl00vzmDk6j/sXl8c5KBGRwasnyZDFzrYd7R53wHJsZhYG\nbgfOBqYDV5jZ9Nhz3P1f3H2uu88FbgX+ejDB97UFSytICRnnzBqV7FBERGQwGjkbMKh4+5Auv3T+\nGFZW7mT55rr4xiUiMkj1JBlaZ2ZfMbPU6PJVYF0PrjsaWOPu69y9FbgPuHA/518B3NuD+yZFJOI8\nurSS4ycXU5idluxwRERkMMrIg+LJUHHwLUMAF8wpJS0lxAMqpCAi0iM9SYa+CBwLbAbKgWOAa3tw\n3Wgg9q9xeXTfXsxsHDABeHYfx681s8Vmtri6+uD7UcfDWxt3sLm2SVXkREQksUrnHXLLUEFWGmfO\nGMnfllTQ3NYR58BERAafnky6WuXul7v7cHcf4e5XuntVnOO4HHjQ3bv9y+3ud7j7fHefX1JSEue3\n7pkFSytITwlxxoyRSXl/EREZIkbNhfoKqN9ySJdfOr+MuqY2nlq5Nc6BiYgMPj2ZZyjDzL5kZr80\nszs7lx7cezMwJma7LLqvO5fTj7vItXdEeGxZJadNG0FO+gGHS4mISB8zs0lmlh5dPynavbsg2XEd\nks4iCofYVe64ScWMLsjkfnWVExE5oJ50k/sDMBI4E3ieIKmp78F1i4DJZjbBzNIIEp4FXU8ys8OB\nYcCrPQ26r728toaaxlbOVxc5EZH+6i9Ah5kdBtxB8GXcn5Ib0iEaOQssdMhd5UIh4+Ijy3hpzTY2\n1zbFOTgRkcGlJ8nQYe7+b0Cju98NnEswbmi/3L0duA5YCKwC7nf3FWb2fTO7IObUy4H7YivW9TcL\nllSQm57CSVOT00VPREQOKBJ97lwE3Oru3wAGZunP9BwonnpI5bU7XXxkGe7wlzdVZltEZH960uer\nLfpaa2YzgS3A8J7c3N0fBx7vsu+mLts39+ReydLc1sHCFVs4e+ZIMlLDyQ5HRES612ZmVwCfAc6P\n7ktNYjy9UzoP1j4D7mB20JePKcziuMOKuH/xJq47+TBCoYO/h4jIUNCTlqE7zGwY8B2Cbm4rgR8n\nNKp+5LnVVTS0tGuiVRGR/u1q4KPAD919vZlNIOjmPTCVzoOGrVBfeci3uHT+GMp3NPHaupo4BiYi\nMrjsNxkysxCw0913uPsL7j4xWlXu130UX9ItWFpBcU4aH51YlOxQRERkH9x9pbt/xd3vjX6Bl+vu\nB/zizszOMrN3zWyNmd3YzfH/MrMl0eU9M6tNyA/QVenc4PUQxw0BnDljJLkZKSqkICKyH/tNhtw9\nAvxrH8XS79Q3t/HM6irOnTWKlHBPGtFERCQZzOwfZpZnZoXAW8BvzOw/D3BNGLgdOBuYDlxhZtNj\nz3H3f3H3ue4+F7gV+GtifoIuRswEC/cqGcpIDXPh3FKeWL6Fuqa2A18gIjIE9eQT/tNmdoOZjTGz\nws4l4ZH1A0+u2Epre4QL5nY7V6yIiPQf+e6+E/gEcI+7HwOcdoBrjgbWuPs6d28F7gMu3M/5V9BX\n00CkZcHwaYdcXrvTpfPH0NIe4ZGlFXEKTERkcOlJMnQZ8CXgBeDN6LI4kUH1Fw8vraBsWCZHjB2Y\nU1WIiAwhKWY2CrgUeLSH14wGYvuQlUf37cXMxgETgGf3cfxaM1tsZourq6t7HvX+lM4NWoZ6UWx1\n1uh8Dh+ZywPqKici0q0DJkPuPqGbZWJfBJdMNQ0tvLxmG+fPKcUOoZKPiIj0qe8TTOWw1t0XmdlE\n4P043v9y4EF37+juoLvf4e7z3X1+SUmcpmEonQe7tkHdoZfHNjMunT+GpeV1rN6yMz5xiYgMIgdM\nhszs090tfRFcMj27uoqOiHPe7IE5TYWIyFDi7g+4+2x3/+fo9jp3/+QBLttMMDlrp7Lovu5cTl91\nkes0al7w2otxQwAfnzea1LDxwGLNOSQi0lVPuskdFbOcANwMXLC/CwaDD2p2EQ4ZU0fkJjsUERE5\nADMrM7OHzKwquvzFzMoOcNkiYLKZTTCzNIKEZ0E39z4cGAa8Gv/I92PEDAil9DoZKsxO4/TpI3jo\n7c20tkfiFJyIyODQk25yX45ZvgAcAeQkPrTkqqhrYkRuuqrIiYgMDHcRJDKl0eWR6L59cvd24DqC\n7nWrgPvdfYWZfd/MYr/0uxy4z70Xg3cORWoGDJ/e62QI4JL5Y9je2MpDb6t1SEQkVsohXNNIMIh0\nUKuobaK0IDPZYYiISM+UuHts8vN7M/vagS5y98eBx7vsu6nL9s1xifBQlM6DVQuCIgq9GL964uQS\njh5fyA8fW8VJU4czIi8jjkGKiAxcPRkz9IiZLYgujwLvAg8lPrTkqqxrZpSSIRGRgaLGzK4ys3B0\nuQqoSXZQvVY6D5p2QO0HvbpNKGT8+OLZtLRH+PZf36GvG7lERPqrnrQM/SxmvR34wN0HdTt7JOJU\n1jZz1kx9cyYiMkB8jmBS1P8CHHgF+GwyA4qL0pgiCsPG9+pWE4qz+caZU/nBY6v425LNXDTvQEOq\nREQGv54MiNkIvO7uz7v7ywTfvo1PaFRJtq2xhdaOCKPVMiQiMiC4+wfufoG7l7j7cHf/OHCganL9\n3/DpEE6Ly7ghgKuPm8ARYwu4ecFKquqb43JPEZGBrCfJ0ANAbPmZjui+QauyNnhAjMpXMiQiMoBd\nn+wAei0lLagqF6dkKBwyfnrJHJraOvjOQ8vVXU5EhryeJEMp7t7auRFdT0tcSMlXUdsEQGmBusmJ\niAxgg2PG7NJ5ULE0KKIQB5NKcvj66VN4cuVWHllWGZd7iogMVD1JhqpjS4ya2YXAtsSFlHwVdUHL\nUKlahkREBrLB0exROg9a6mD7urjd8poTJjJ3TAHffXg51fUtcbuviMhA05Nk6IvAt81so5ltBL4J\n/FNiw0quitomMlPDFGSlJjsUERHZDzOrN7Od3Sz1BPMNDXyxRRTiJBwyfnbJbBpbO7jp4eVxu6+I\nyEDTk0lX17r7R4DpwHR3P9bd1yQ+tOQJ5hjKwHoxp4OIiCSeu+e6e143S667H8pcev1PyeEQTo9r\nMgRw2PBcvnbaZJ5YvoXH1F1ORIaonswz9P/MrMDdG9y9wcyGmdkP+iK4ZKmoa9aEqyIi0j+EU2Hk\nLKhYEvdbX3vCRGaX5fNvDy+npkHd5URk6OlJN7mz3b22c8PddwDnJC6k5KuobdJ4IRER6T9K50Hl\nUohEDnzuQUgJh/jpxXOob27juwtWxPXeIiIDQU+SobCZpXdumFkmkL6f8we0lvYOqutbGKVKciIi\n0l+UzoPWeti+Nu63njoyl6+eOplHl1XyxDvqLiciQ0tPkqH/BZ4xs8+b2TXAU8DdiQ0rebbWBd0E\n1E1ORET6jQQUUYj1TydOYuboPP7t4eVsb2w98AUiIoNETwoo/Bj4ATANmAosBMYlOK6kqaiLzjGk\nbnIiItJfFE+B1KyEJUOp0e5ydU1tfO8RdZcTkaGjJy1DAFsJ5mu4BDgFWJWwiJJME66KiEi/E06J\nFlFITDIEMG1UHtedPJmHl1SwcMWWhL2PiEh/ss9kyMymmNl3zWw1cCuwETB3P9ndb+uzCPvYnmRI\nLUMiItKP7C6i0JGwt/j/Tp7EtFF5/N+HlrND3eVEZAjYX8vQaoJWoPPc/Xh3vxVI3F/gfqKirpnC\n7DQyUsPJDkVERGSP0nnQtgu2vZ+wt0gNh/jZJbOp3dXKxb96hZUVOxP2XiIi/cH+kqFPAJXAc2b2\nGzM7FRj0s5B2TrgqIiLSryS4iEKnGaX53PO5o9nZ3M7Hf/kyf3h1A+6e0PcUEUmWfSZD7v43d78c\nOBx4DvgaMNzM/sfMzuirAPtaZW0zo1Q8QURkSDCzs8zsXTNbY2Y37uOcS81spZmtMLM/9XWMuxUd\nBmk5CU+GAI49rJgnvnoCx04q4t8eXsEX//gmtbvUbU5EBp+eVJNrdPc/ufv5QBnwNvDNhEeWJBW1\nTYzWeCERkUHPzMLA7cDZwHTgCjOb3uWcycC3gOPcfQbBF4PJEQrDyNl9kgwBFOekc+dnjuL/njON\nZ1ZVcc4vXmTxhu198t4iIn2lp9XkAHD3He5+h7ufmqiAkmlncxv1Le3qJiciMjQcDaxx93Xu3grc\nB1zY5ZwvALe7+w4Ad6/q4xg/rHQebHkHOtr75O1CIeMLH5vIX/75WFLCIS674zVue/Z9OiLqNici\ng8NBJUODXWVtM4C6yYmIDA2jgU0x2+XRfbGmAFPM7GUze83MzuruRmZ2rZktNrPF1dXVCQqXIBlq\nb4Jt7ybuPboxZ0wBj33leM6ZNYqfPfken/rd62zd2dynMYiIJIKSoRgqqy0iIl2kAJOBk4ArgN+Y\nWUHXk6K9Jua7+/ySkpLERdNHRRS6k5uRyn9fPpeffHI2b23cwdm/eJHn3k1uQ5mISG8pGYpRUacJ\nV0VEhpDNwJiY7bLovljlwAJ3b3P39cB7BMlRchROhPS8pCRDAGbGpUeN4dEvH8/w3HSuvmsRP3xs\nJa3tkaTEIyLSW0qGYlTUNhEOGcNzlQyJiAwBi4DJZjbBzNKAy4EFXc75G0GrEGZWTNBtbl1fBvkh\noRCMmpO0ZKjTYcNz+duXjuNTHxnHb15czzn//SLPv5fA7oEiIgmiZChGRW0zI/MyCIcG/XRKIiJD\nnru3A9cBC4FVwP3uvsLMvm9mF0RPWwjUmNlKgmkmvuHuNcmJOKp0LmxZDh1tSQ0jIzXMv398Jnd+\ndj5tHRE+c+cbfP73i1hX3ZDUuEREDkZCk6EBNX8DmnBVRGSocffH3X2Ku09y9x9G993k7gui6+7u\n17v7dHef5e73JTdignFDHS1QtSrZkQBwyuEjePJfPsa3zj6c19dv58yfv8APH1vJzubkJmsiIj2R\nsGRowM3fQDBmSMUTRESkX0tiEYV9SU8J808nTuLZG07kE/PK+O1L6zn5p//gT69vVBluEenXEtky\nNKDmb4hEnC11zSqrLSIi/duwCZCR36+SoU7DczP48cWzeeS645lYks23H3qH8259idfWJbdnoYjI\nviQyGRpQ8zdsa2ihrcMZrW5yIiLSn5kFrUP9MBnqNHN0Pvf/00e59Yp57Gxq4/I7XuP/+9832bR9\nV7JDExH5kGQXUOg38zdsjs4xpJYhERHp90bNha0roL0l2ZHsk5lx/pxSnvn6iVx/+hSeXV3Fqf/5\nPD98bOXuef1ERJItkcnQgJq/obIumElbY4ZERKTfK50HkTaoWpnsSA4oIzXMV06dzHM3nMR5s0bx\nu5fWc8JPnuMr977NsvLaZIcnIkNcIpOhATV/Q+e3VKOVDImISH/XD4soHMio/Ez+87K5PP+Nk7n6\n2PE8u7qKC257mUt/9SoLV2xRoQURSYqEJUMDbf6GitpmstLC5GWmJOPtRUREeq5gLGQWDqhkqNOY\nwiy+c950Xv3WKXzn3Glsrm3in/7wJqf8xz+4+5UN7GptT3aIIjKEJPSTv7s/DjzeZd9NMesOXB9d\nkiqYYygTM024KiIi/dwAKKJwILkZqVxzwkQ+e+x4Fq7Yym9fWsd3F6zgP596jyuPGctnPjqekfkq\naiQiiaVmkKiKuiZG6Y+uiIgMFKVz4eVfQFsTpA7cLt4p4RDnzh7FubNH8eYHO7jzpfX8+vm1/OaF\ndZw5cyQXzCnlxCklZKSGkx2qiAxCSoaiKmqbmT4qL9lhiIiI9Mzo+RBph/cWwoyPJzuauDhy3DCO\nHDeMTdt3cdfLG/jbks08tqyS3PQUTp8xgvNnl3L85GJSw8kuhisig4WSIaClvYNtDS2qJCciIgPH\n5DNg+Ax48jvBelpWsiOKmzGFWdx0/nS+fc7hvLK2hkeWVrBwxRb++tZmCrJSOWvGSM6bXcpHJhaS\nosRIRHpByRCwJVpWW93kRERkwAinwDk/hd+fAy/+B5z6b8mOKO5SwiE+NqWEj00p4YcXzeLF96t5\nZGkFjyyt4L5FmyjOSePsmaM4b/YojhpfSCikcb8icnCUDLFnwlWV1RYRkQFl/HEw+zJ45b9h7pVQ\nNCnZESVMWkqIU6eN4NRpI2hu6+Af71bxyNJKHnhzE3947QNKctM5YXIxJ04p4fjDiinKSU92yCIy\nACgZAiproy1DSoZERGSgOf3f4d0n4PFvwFV/CSrNDXIZqWHOmjmKs2aOorGlnWdWV/Hkii08u7qK\nv74VzO8+c3QeH5tcwgmTSzhy3DDSUtSdTkT2pmSIPROuqpuciIgMOLkj4ORvw99vhNWPwrTzkx1R\nn8pOT+GCOaVcMKeUjoizfHMdL75fzQvvbeOOF9bxy3+sJTstzEcnFXHC5KDL3fiiLE2lISKAkiEg\nKKtdlJ2msp0iIjIwHfUFeOsP8PdvwaRTB1UxhYMRDhlzxhQwZ0wB150ymfrmNl5dW8ML0eTo6VVV\nAJQNy+S4ScUce1gRx04qpiRXXepEhiolQwRltVVJTkRk6DGzs4BfAGHgt+5+S5fjnwV+CmyO7rrN\n3X/bp0H2RDgFzv0Z3HX2oC2mcChyM1I5Y8ZIzpgxEoAPahp54b1qXnx/G48vr+TPizcBMHVELsce\nVsRxk4o5emIheRmpyQxbRPqQkiGCbnITS7KTHYaIiPQhMwsDtwOnA+XAIjNb4O4ru5z6Z3e/rs8D\nPFjjjoXZlwfFFOZcAcWHJTuifmdcUTaf+mg2n/ro+N1d6l5eu41X1tTwp9c3ctfLGwiHjFmj8zku\nmhwdMW6Yeo6IDGJDPhlyd/7/9u48yq6qzPv497m37lDznFRSVRlIwhAIhFigII0yiAFfobtBQVob\nUFbUloZXRUW7GxDt7kj72q3A0qYxLFi20Coi0DIICgqiQIQQQkLIIJBKKlMlNdetcb9/7FNDKpW5\nqm7de36ftc46Q51b9ey6yd15svd59pamTt47tyLdoYiIyMQ6FVjvnNsIYGb3AxcBI5OhzPGBW2Dt\no/DYl+DjPw9FMYXDNXxK3d+9fy6pnj5eeaeJ5zfs5Pfrd/KD327kjqc3EM+JsLCmhLpZpZwyq4xF\nM0spztXIkUi2CH0y1JLqpb27T2W1RUTCpxrYNOy8Hnj3KPddbGZnAm8Cn3fObRrlnsmhcCqc9Q/w\n+FdgzSMw/8J0R5QxklwALq0AACAASURBVDFfZOG0OeV88bxjaE318OKfd/GHDY289PbuwWIMZn5a\n3UByVDerTP+GEMlgoU+GBivJlaiSnIiI7OUR4D7nXJeZfRq4Bzh75E1mtgRYAjBjxoyJjXCkU66G\nV4JiCnPPgbimgR+OwmRscF0jgI7uXlZsamL5W7t56a1d/OKVLfzoj+8AML04Sd2sMupmlbKwtoRj\nqgpJ5GhqnUgmCH0y1NDskyEVUBARCZ3NQO2w8xqGCiUA4JxrHHZ6F3DraN/IOXcncCdAXV2dG9sw\nD1E0By74t2HFFG5MazjZIi+ew+lzKjh9jp9W39fveGNry2By9MKfG3n41S0AxKLGsVVFLKgp5sTq\nYhbUFHP01EJiUa11JDLZhD4Z2hwsuKohbhGR0HkJmGdms/FJ0GXA5cNvMLNpzrmG4PRCYM3EhniY\nZp7uiyj8/ntw0uUqpjAOohHj+OnFHD+9mCtOn4Vzjvrdnby2uZmV9c28trmJR17dwo9f8KNH8ZwI\n86cVcWJNMQuqizmptoQ5lQVEI3quSySdQp8MNTR1khMxKgq0xoCISJg453rN7BrgCXxp7WXOudfN\n7BZguXPuYeBaM7sQ6AV2AVemLeBD9YFb4I1fqpjCBDEzasvyqC3L44IF0wDo73e8s6uDlZubea2+\niZX1zTzwp3ru/cPbAOTFoyyoLmZhUMjhpNoSphcntSCsyAQKfTK0pamTquKk/mdGRCSEnHOPAo+O\nuHbjsOOvAl+d6LjGRMEUOPsf4bEvw5qHYf5F6Y4odCIRY1ZFPrMq8rnwpOmAT5A27mxnZZAcrdjU\nxN2/f4vuvn4AKgoSLKwt5qQanxydWFNMSV48nc0QyWpKhppSTC/WFDkREclCdZ+Cl++Fx78Gc89V\nMYVJIBIx5k4pYO6UAv56UQ0A3b39vLG1hVc3NbFiUzOv1jfx6ze244Knz2aV5zF/ehFzKwuYE7x2\nTmWB1j8SGQNKhpo7qZtZmu4wRERExl40By74Nty9GH73bTj3pnRHJKOI50Q4saaEE2tK+MRp/lpr\nqofXNjfz6qZmXt3UxJqGVh5ftZX+IEEyg9rSvMHEaniipHWQRA5eqJOhvn7H1uaUKsmJiEj2mnma\nL6Lw/G1QeyocvVjPD2WAwmRsj+p1AKmePt5qbGf99rY9tufW76S7t3/wvsrCBEdV5HNUZQFzKvM5\nqjKfoyoKqCnNJUcV7UT2EOpkaGdbF739jmlKhkREJJt94BbYvBzuuwxmvw/O+yZMOzHdUckhSsai\nHFtVxLFVRXtc7+t31O/uYN22NtbvaGPD9jY27mzn8VUN7O7oGbwvFjVmlucPJkpHVeYzuyKf6pJc\nphbp+WkJp1AnQ5uDBVerteCqiIhks4JK+OzzsHwZPLMU/vNMWHi5L7BQND3d0ckRikZ8kjOzPJ9z\nmbrH13a3d7NxZxsbdrSzYUcbG4P902u309M3tCRWTsSYVpKkuiSX6pI8qktzqSnJpbo0l+qSXKaV\nJLWQrGSlUCdDW4JkaJoKKIiISLaLxuDdn4YTL4Vnvw0v/Ces+jmc/vfw3usgUZDuCGUclObHeVd+\nGe+aWbbH9d6+fjbt7uStxnY27+5kc1Pn4P7363eyrTU1WMAB/MzKKYUJakrzqC3N9fsyv68pzWV6\nSa4WlZWMFOpkqCFYcFXPDImISGjklvhpcqdcDU99HX53K7x8D5z1D3DyxyGi//0Pg5xohNkVfprc\naLp7+9nanKK+qWMwSarf3Un97g6Wv72bR1Y20Nc/lC1FDKqKkj45KvOjSpWFCcoLElQUJCgviFOR\nn6AoN0frKMmkEupkaHNTJ/nxKEXJUP8aREQkjEpnwUfuhvf8HfzqH+CRa+GFH8B53/BluCXU4jkR\nZpTnMaM8b9Sv9/b109Ccon53J5t2dwwmSvW7OvnjhkYaWvYcWRoQixrl+QkqCuN+X+CPpxUlmVYS\nTMkrTlKWH1fSJBMi1FlAQ3Mn00ty9ZdNRETCq/YU+OQTsPoheOom+NHFMOcc+NC3oeyodEcnk1RO\nNEJtWR61ZXmcRvleX+/t62dXRzeNbd3sbOsa3O9s66axrctfa+9m3bZWdrZ1Dy46OyCRE2F6SS7T\nS5JMK/bT8KYX+4SpMJlDIidCIidKMub3iViEZE6UWNT07zo5JKFOhrY0qay2iIgIZnD8X8Ix58NL\nd8Ez34L/Ohsu/RHMOiPd0UkGyolGmFKYZErhgYtUOedobO+moSnF5qZOGpo72dLUyZbmFFuaOnlu\n3d7PMO2LGSSD5CiRE6E0L05lYYKpRUmmFCb8VpRkalGCKYVJKgsTWrw25EKeDHVyQnXRgW8UEREJ\ng5wEnPY5nxT9+FK49yL4P/8Oi/423ZFJFjMzP12uIMGCmuJR7+np62dbS4qG5hTtXb109faT6umj\nq7ffbwPHPX2kBvY9fnRqe0uK9dvb2NHql1QZqTg3xpQgYZpalKSqOEFVUZKq4lyqipJMLU5QkZ8g\notLjWSm0yVCqp4/G9m6mq5KciIjInsqOgqufgp9eCQ//PexY69cqUnEFSZNYNBJUrhv9GaaD0d/v\n2N3RzbaWLra3ptje2sX2Fr/f1pJiW0sXGzbsZHtr1x7FIcCXHp9SmKCqOElVcZLy/AR5iSgF8Rzy\nEjnkx6PkJ3LIT0TJj+cEx/56XiKHvFhUydQkFdpkqKHZV5LTgqsiIiKjSBbD5T+FJ74Gf7gddq6D\ni++CpGZUSGaKRIzyAl/hbj77/nPc1+9obOuioTnF1pYU21pSbB12/MbWVna3N9Le3Ud3b/8+v89w\nZgRJUpSCRA4FiaGEqXD4cXJoK0rGKEzGhs5zYxTEc5RUjbHwJkPBGkPTteCqiIjI6KI5cMGtUHk0\nPPplWPZB+Nj9UDoz3ZGJjJtoxJhSlGRKUZKTDnBvT18/HV19tHf30tHdS1tXHx1dvbR19dLR7a+3\nd/nrbanguLt38HhXewdtXf64vatvr0ISI5lBQXwoOSrJi1GWH6c0L9jy45TlxyjJi1OWF/dfy4+T\nH4+qsMQ+hDYZ2hwkQ9UaGRKRQ9DT00N9fT2pVCrdoUyIZDJJTU0NsVgs3aFIOp1yNZTNgZ9e4Qsr\nXPbfMOM96Y5KJO1i0QjFeRGK88bmM7Krt4/WVC8tnT20pnr9caqH1lTP4PWW4FpLZy9NHd2s3drK\n7o4emjq6GeWRqCBOoyQvTmlejJLcOCV5MUrz4pTkx4JEKkZxrt+X5seJRiwoWOFwDhwE++B82HFB\nIofygjgFicxcQyq0ydCWYMHVqmKNDInIwauvr6ewsJBZs2Zl5If+oXDO0djYSH19PbNnz053OJJu\nc86Cq3/tCyvc82G48DY46bJ0RyWSVRI5URIFUSoKEof82v5+R0uqh13t3ezu6GF3eze7O/y2q90n\nS7s7umnq6OHtxg5WbGqiqaPngKNRBysejVBeEPdbfrDQbkGCsvw45fnxwSIZlYUJKgri5EQjY/Jz\nj1Rok6GG5k4qChIkcvQwqIgcvFQqFYpECHyFp/Lycnbs2JHuUGSyqJjnCyv85G/hwU/7wgpn/xNE\nJsc/akTCLBLxoz8lefGDfo1zjo7uPpo6ffLU1NFDU2f3YAEJM8Pw0/MMC/bB9aAbbO/q9etItXex\nq62bxna/ltT67W00tneR6tk72TJjMEGaEpQ9rxwofR6UPJ9alGBmef4Y/Gb2L7TJ0OamTj0vJCKH\nJQyJ0IBsb6uZLQa+C0SBu5xzS/dx38XAz4BTnHPLJzDEySevDD7xIPzyi/Dcd2Dnm3DBv0FBlZIi\nkQxjZoPFG8bj0ZGBZGtXezc72rrY2drF9tYudgzuU+xo7WLdtta9Sp9PLUrwwtfOHfOYRhrXZOhA\nnYyZXQn8G7A5uHS7c+6u8YxpQENzirmVBRPxo0REZBIysyhwB/ABoB54ycweds6tHnFfIXAd8MLE\nRzlJRWPw4e/ClON8tbk3/heiCSiuhuJav5XUQnHN0HFRtV/HSERCY3iyVVu2/7Lo/f2Ops4eX/a8\npeugK/UdqXFLhg62kwH+xzl3zXjFMRrnHFuaOjlzXuVE/lgRkSPW2NjIOeecA8DWrVuJRqNUVvrP\nshdffJF4/MDTI6666ipuuOEGjjnmmHGNNQOcCqx3zm0EMLP7gYuAkf3UN4BvAV+a2PAmOTN4z2dh\nxmlQ/xI0b4KmTX6/4dfQuhX/2PXgC6BgKpTPgaknwNTjoeoEqDwO4oe/doyIZIdIxCjL9xXwjq2a\nuJ87niNDB9vJTLiWTl/uUNPkRCTTlJeXs2LFCgBuvvlmCgoKuP766/e4xzmHc47IPqYs3X333eMe\nZ4aoBjYNO68H3j38BjNbBNQ6535pZvtMhsxsCbAEYMaMGeMQ6iQ2faHfRurtgpbN0Fw/lCQ1bfLT\n6l75EfS0+/ss4ivVTT3eJ0lVQaJUXAtZPk1TRNJvPJOhA3YygYvN7EzgTeDzzrlNI28Y605m8+Aa\nQyqrLSKH7+uPvM7qLS1j+j3nTy/ipg8ff8ivW79+PRdeeCEnn3wyr7zyCk8++SRf//rXefnll+ns\n7OTSSy/lxhtvBOCMM87g9ttv54QTTqCiooLPfOYzPPbYY+Tl5fHQQw8xZcqUMW1TpjKzCPAd4MoD\n3eucuxO4E6Curm4fxW1DJicBZUf5baT+fmh6C7a9DltXwbZV0LACVv9i6J5EMcx4Nyy8HI65QFPs\nRGRcpLuAwiPAfc65LjP7NHAPcPbIm8a6k9kSJEPTVFZbRLLIG2+8wb333ktdXR0AS5cupaysjN7e\nXs466ywuueQS5s+fv8drmpubed/73sfSpUv5whe+wLJly7jhhhvSEX46bAZqh53XMPQMK0AhcALw\nTFBIogp42MwuDH0RhSMViQwlSsd9eOh6VytsW+2To22r4M1fwU+vhLxyOPEyWPQJ/5ySiMgYGc9k\n6ECdDM65xmGndwG3jmM8gxqateCqiBy5wxnBGU9z5swZTIQA7rvvPn74wx/S29vLli1bWL169V7J\nUG5uLueffz4A73rXu3j22WcnNOY0ewmYZ2az8f3TZcDlA190zjUDFQPnZvYMcL0SoXGUKPSjQTOC\niST9fbDhaXjlXnjxTvjjHVBzCpz8CTjhr/39IiJHYDxrYA52MmYWx3cyDw+/wcymDTu9EFgzjvEM\n2tyUIha1w1rQSkRkssrPH1qPYd26dXz3u9/lN7/5DStXrmTx4sWkUqm9XjO84EI0GqW3t3dCYp0M\nnHO9wDXAE/j+5yfOudfN7BYzuzC90QkAkSjMOxc+ei988Q0475/96NEj18K3j4GHPgfvvABOMxNF\n5PCM28iQc67XzAY6mSiwbKCTAZY75x4Grg06nF5gFwcxL3ssNDR3UlWcJBLRg5kikp1aWlooLCyk\nqKiIhoYGnnjiCRYvXpzusCYd59yjwKMjrt24j3vfPxExyT7kV8Dp18Bpn4P65fDyPbDq574YQ8Ux\n8K4r/abKdCJyCMb1maEDdTLOua8CXx3PGEazpamT6cWaIici2WvRokXMnz+fY489lpkzZ/Le9743\n3SGJjA0zqD3Fb4uXwusPwsv3whNfhef+Hc74PNRdBTH18yJyYOYybGi5rq7OLV9+ZNO137v0N5w6\nu4x/v3SUUqAiIvuxZs0ajjsuXA9wj9ZmM/uTc65uHy8JtbHop+QwvPNHePpf4M+/hYIq+IsvwKIr\nIHYExZL6+6DpHSiZ6Ys+iEjGONh+Kt3V5CZcX79ja0tKawyJiIhkkxnvgSsehree80nRY1+G5/4D\nzvyiL7hwsKW5u1phw29g7eOw7gnoaISial+w4YRLYNpJWv9IJIuELhna3pqir98xTdPkREREss+s\nM+DKX8Kff+eTol9+MUiKroeFfwPR2N6vaXrHJz9vPuaTqb5uSJbAvA9AdR1sfBr++H14/jYonwcL\nLvGJUcXcQ4+vbYdfU2nrSkg1+9Gn/t4RWx/09ex5XjEPTrhYyZjIGAtdMrSlyVdTUlltERGRLGUG\nR70PZp/pR3me/hd45Dp49jvwvi/Dgo9Cw6s++Vn7OGx/3b+ufC6cugSOOR9q3wPR4J9J7/kMdOyC\n1Q/BqgfgmaXwzL/6xOSES3ySUly9dxyt23zis2WF3ze8Ci3DVhmJJiCS439OZPgWhUhs6NwM1j8J\nz3/Px3jCJT4hq5g3/r9LkSwXwmTIrzE0XcmQiIhIdjODuefAnLNh/VPw9D/7ctz/+3k/+mNRmHEa\nnPdNOPr8/Y/05JX5wgx1V0HLFl/JbtXP4Ml/gidvhJmnw3EXQufuoQSobetAID6JmXk6TFsI0xdC\n1YmQLDr4tgxPxn77LfjtUqhaMJSMldQe+HuIyF5CmwxN0zNDIiIi4WDmp7zNPRfefBzWPemToLnn\n+CTnUBVN92W+T78GGjf4BOW1n8LjXwEMKo72I1ODic+CI18gdo9krMFX0Vv1M3jqJr/VvtsnRsf/\nJRRMObKfJRIioUuGGppTFCZyKEqOMmdYREREspeZnwJ3zPlj9z3L5/ipd2d+CZrehrwKSBSM3fcf\nTdE0OO3v/LZrYzBK9QA89iWfkJXOAtcP/f3g+vwzR3vsh13PK/OJ4tHnw1Hv1zpNEjqhS4Y2N3Vq\nVEhEMlZjYyPnnHMOAFu3biUajVJZWQnAiy++SDweP6jvs2zZMi644AKqqqrGLVaRUDHzSchEKzvK\nF4c483rYttonRbs2Dj17ZFFfFtyiw86jYBG/b3oHXv+FX6spJ+kTomPOh6MXQ6E+HyT7hS4Zamju\n1PNCIpKxysvLWbFiBQA333wzBQUFXH/99Yf8fZYtW8aiRYuUDIlkk6nz/Xaoervh7d/7KYRrH/N7\ngOkn+xGjY873U/1UxU6yUOiSoS1NKRZUl6Q7DBHJBo/dAFtfG9vvWbUAzl96WC+95557uOOOO+ju\n7ub000/n9ttvp7+/n6uuuooVK1bgnGPJkiVMnTqVFStWcOmll5Kbm3tII0oikoVy4jDnLL8tXgrb\n1wxV2nvmX+GZf4GiGpjzfiiZ5Z+ZKq726y8VTYd4frpbIHLYQpUMdXb3sau9m2pNkxORLLNq1Soe\nfPBBnn/+eXJycliyZAn3338/c+bMYefOnbz2mk/ampqaKCkp4bbbbuP2229n4cKFaY5cRCYVs6ER\npr/4ol8Xad0TfsRo7ePQsXPv1ySLhxKjoiBJKpwKsTy/2G004feDW3LPa5Ec6GrxC9x27IbOXb56\nXueu4NquYdeaABdM8xtZjnzkcRQYPprl9ozbjTiPF/jiE4VVfl8wddg25cAL9/Z2+bWjOpv8PtUM\nqSboTfk2x/IglrvnPj7sOBrX6FsahCoZamhWWW0RGUOHOYIzHp566ileeukl6urqAOjs7KS2tpYP\nfvCDrF27lmuvvZYPfehDnHfeeWmOVEQySkElnPxxvwH0pKC1wZcXb9kcbFv81lwPDSuhffsY/XCD\n3BLILYO8cp9sTT3eJ0J7LVY74tz1Q3fXKN9yZLIxcO6gfSdsemH0hA/8QrwDiVJOcvSk54iaG4FE\nka9GOOVYqDwOpgRbwVQlSuMkVMnQwIKr04qVDIlIdnHO8clPfpJvfOMbe31t5cqVPPbYY9xxxx08\n8MAD3HnnnWmIUESyQiwJZbP9ti+93T4h6u3yCUJv19BxX/fe1/p7fRKQFyQ9uWX+OFkcjO5MsL4e\naN8BbdugbXuw3+YX0R24lmrxiVrRNB9nsiTYF0Nu6dBxssRPQ+ztgp4O6Okctu/c+1rHLtixFtb8\nry9qMSBZAlPm75kkVRztKxdGE0MLBMshC9VvbkswMlStkSERyTLnnnsul1xyCddddx0VFRU0NjbS\n3t5Obm4uyWSSj3zkI8ybN4+rr74agMLCQlpbW9MctYhkpZw4FNekO4rDF40FU/6mpy8G53xCtn0N\n7HgDtq+G7W/4aoGp5r3vt6gfrRo5HXFgmmIs6ZO0vHJf/j2/IjguD46D85yDfH7UuaHRuK5Wv9jw\nHtuuUa7t9q9JFPopiYmCYL+P89xSmPHusf29jiJcyVCw4OrU4gPM+RQRyTALFizgpptu4txzz6W/\nv59YLMYPfvADotEon/rUp3DOYWZ861vfAuCqq67i6quvVgEFEZHJyCx4bmmKX8B3gHN+muL2Nb6E\nek/HnqNsI/cDI3E9nf417Tt9UjLy+akBAyN0FoX+Hp+89PWMOO715wdsQ8SPaOWWBolYhR/p62qD\ntq3Q2Abdbf68p33v1xfVwBdeP6xf36EwN/LhsUmurq7OLV++/LBe++Ar9fx27Q7+47KTxzgqEQmL\nNWvWcNxxx6U7jAk1WpvN7E/Oubo0hTSpHUk/JSIy7vr7fELUvtM/H9XRGBw3Dm2uPyhCEfNT8EY9\njvnkJl4YJDulQ4lPbikkiv0aVwcbU3d7kBy1+gTJ9UPtKYfdzIPtp0I1MvRXJ9fwVydn8LCtiIiM\nKTNbDHwXiAJ3OeeWjvj6Z4DPAX1AG7DEObd6wgMVERkrkaifGpdfke5IhkSikCzy20T/6An/iSIi\nIpOAmUWBO4DzgfnAx8xs5IqVP3bOLXDOLQRuBb4zwWGKiMg4UjIkInKIMm168ZHI8raeCqx3zm10\nznUD9wMXDb/BOdcy7DSffU60FxGRTKRkSETkECSTSRobG7M9SQB8ItTY2EgymbULVVcDm4ad1wfX\n9mBmnzOzDfiRoWtH+0ZmtsTMlpvZ8h07doxLsCIiMvZC9cyQiMiRqqmpob6+nrD8gzeZTFJTE+5n\nLZ1zdwB3mNnlwD8CV4xyz53AneALKExshCIicriUDImIHIJYLMbs2ftZbFAyyWagdth5TXBtX+4H\nvj+uEYmIyITSNDkREQmrl4B5ZjbbzOLAZcDDw28ws3nDTj8ErJvA+EREZJxpZEhERELJOddrZtcA\nT+BLay9zzr1uZrcAy51zDwPXmNm5QA+wm1GmyImISOZSMiQiIqHlnHsUeHTEtRuHHV834UGJiMiE\nsUyriGRmO4C3j+BbVAA7xyicyUjty1zZ3DbI7vZlc9tg9PbNdM5VpiOYyU791AGpfZkrm9sGal8m\nO+x+KuOSoSNlZsudc3XpjmO8qH2ZK5vbBtndvmxuG2R/+yabbP99q32ZK5vbBmpfJjuStqmAgoiI\niIiIhJKSIRERERERCaUwJkN3pjuAcab2Za5sbhtkd/uyuW2Q/e2bbLL99632Za5sbhuofZnssNsW\numeGREREREREIJwjQyIiIiIiIkqGREREREQknEKVDJnZYjNba2brzeyGdMczlszsLTN7zcxWmNny\ndMdzpMxsmZltN7NVw66VmdmTZrYu2JemM8YjsY/23Wxmm4P3cIWZXZDOGA+XmdWa2dNmttrMXjez\n64LrWfH+7ad92fL+Jc3sRTN7NWjf14Prs83sheDz83/MLJ7uWLNRNvdTkF19lfqpjP6cUz+V2e/f\nmPZToXlmyMyiwJvAB4B64CXgY8651WkNbIyY2VtAnXMuKxbTMrMzgTbgXufcCcG1W4FdzrmlwT8S\nSp1zX0lnnIdrH+27GWhzzn07nbEdKTObBkxzzr1sZoXAn4C/BK4kC96//bTvo2TH+2dAvnOuzcxi\nwHPAdcAXgJ875+43sx8Arzrnvp/OWLNNtvdTkF19lfqpzKV+KuPfvzHtp8I0MnQqsN45t9E51w3c\nD1yU5phkH5xzvwN2jbh8EXBPcHwP/i92RtpH+7KCc67BOfdycNwKrAGqyZL3bz/tywrOawtOY8Hm\ngLOBnwXXM/b9m+TUT2UQ9VOZS/1UZhvrfipMyVA1sGnYeT1Z9AcD/4fgV2b2JzNbku5gxslU51xD\ncLwVmJrOYMbJNWa2MpiekJHD88OZ2SzgZOAFsvD9G9E+yJL3z8yiZrYC2A48CWwAmpxzvcEt2fb5\nOVlkez8F2d9XZd3n3Ciy4nNugPqpzDSW/VSYkqFsd4ZzbhFwPvC5YHg7azk/vzPb5nh+H5gDLAQa\ngP+X3nCOjJkVAA8A/9c51zL8a9nw/o3Svqx5/5xzfc65hUANfrTi2DSHJNkjNH1VNnzOjSJrPudA\n/RQZ/P6NZT8VpmRoM1A77LwmuJYVnHObg/124EH8H4xssy2YBzswH3Z7muMZU865bcFf7n7gv8jg\n9zCYw/sA8N/OuZ8Hl7Pm/Rutfdn0/g1wzjUBTwOnASVmlhN8Kas+PyeRrO6nIBR9VdZ8zo0mmz7n\n1E9l9vs3YCz6qTAlQy8B84JKE3HgMuDhNMc0JswsP3hADjPLB84DVu3/VRnpYeCK4PgK4KE0xjLm\nBj6AA39Fhr6HwYONPwTWOOe+M+xLWfH+7at9WfT+VZpZSXCci3+Yfw2+s7kkuC1j379JLmv7KQhN\nX5UVn3P7kkWfc+qnMvv9G9N+KjTV5ACCEoL/AUSBZc65f05zSGPCzI7C/w8bQA7w40xvm5ndB7wf\nqAC2ATcBvwB+AswA3gY+6pzLyIc799G+9+OHrh3wFvDpYXOXM4aZnQE8C7wG9AeXv4afr5zx799+\n2vcxsuP9OxH/4GkU/x9mP3HO3RJ8ztwPlAGvAB93znWlL9LslK39FGRfX6V+KqM/59RPZfb7N6b9\nVKiSIRERERERkQFhmiYnIiIiIiIySMmQiIiIiIiEkpIhEREREREJJSVDIiIiIiISSkqGREREREQk\nlJQMiRwmM+szsxXDthvG8HvPMrOMrP8vIiKTh/oqkf3LOfAtIrIPnc65hekOQkREZD/UV4nsh0aG\nRMaYmb1lZrea2Wtm9qKZzQ2uzzKz35jZSjP7tZnNCK5PNbMHzezVYDs9+FZRM/svM3vdzH4VrLIs\nIiJyxNRXiXhKhkQOX+6IqQeXDvtas3NuAXA7fjV5gNuAe5xzJwL/DXwvuP494LfOuZOARcDrwfV5\nwB3OueOBJuDicW6PiIhkH/VVIvthzrl0xyCSkcyszTlXMMr1t4CznXMbzSwGbHXOlZvZTmCac64n\nuN7gnKswsx1AZolozQAAAQBJREFUjXOua9j3mAU86ZybF5x/BYg55745/i0TEZFsob5KZP80MiQy\nPtw+jg9F17DjPvSMn4iIjC31VRJ6SoZExselw/Z/CI6fBy4Ljv8GeDY4/jXwWQAzi5pZ8UQFKSIi\noaa+SkJP2bvI4cs1sxXDzh93zg2ULC01s5X4/zH7WHDt74G7zexLwA7gquD6dcCdZvYp/P+qfRZo\nGPfoRUQkDNRXieyHnhkSGWPBPOw659zOdMciIiIyGvVVIp6myYmIiIiISChpZEhEREREREJJI0Mi\nIiIiIhJKSoZERERERCSUlAyJiIiIiEgoKRkSEREREZFQUjIkIiIiIiKh9P8BDHcMgTTc5hgAAAAA\nSUVORK5CYII=\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "The confusion matrix:\n", "\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0cAAAEYCAYAAACX/Y2IAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkyMzMzMzMzPAyZGZmZmZmRkA/w86BuMDGFGYjAAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "\n", "The classification report:\n", "\n", " precision recall f1-score support\n", "\n", " Neutral 0.89 0.94 0.92 3500\n", " Happy 0.92 0.88 0.90 3500\n", " Sad 0.89 0.89 0.89 3205\n", "\n", " accuracy 0.90 10205\n", " macro avg 0.90 0.90 0.90 10205\n", "weighted avg 0.90 0.90 0.90 10205\n", "\n" ] } ], "source": [ "# Training the model (3 classes):\n", "start_time = time.time()\n", "history = model.fit(X_train, y_train, validation_data = (X_val, y_val), \n", " batch_size = mini_batch_size, \n", " epochs = num_epochs, verbose = 2)\n", "end_time = time.time()\n", "\n", "# Creating a new directory to save model and history:\n", "path = root_dir + '/Models/' + str(model_number)\n", "if not os.path.exists(path):\n", " os.mkdir(path)\n", "\n", "model.save(path + os.sep + 'model.h5') # Saving the model\n", "json.dump(history.history, open(path + os.sep + 'Model_history.json', 'w')) # Saving the history\n", "plot_model(model, to_file=path + os.sep + 'Model.png', show_shapes = True) # Saving the model visualization\n", "\n", "# Updating the Excel sheet:\n", "row_index = params_sheet[params_sheet['model'] == model_number].index[0]\n", "params_sheet.loc[row_index, 'train_acc'] = round(max(history.history['acc'])*100, 2)\n", "params_sheet.loc[row_index, 'val_acc'] = round(max(history.history['val_acc'])*100, 2)\n", "params_sheet.loc[row_index, 'train_time'] = round((end_time - start_time)/60)\n", "params_sheet.to_excel(root_dir + '/Model Params.xlsx')\n", "\n", "# Plotting training history:\n", "summary_visualizer(history, path, X_test, y_test, params_sheet, model_number)" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "QgSsW3APEApi" }, "source": [ "#### 1.3.2. Model with 5-Classes (Happy, Sad, Surprised, Angry and Neutral)\n", "\n", "Similarly, for 5 classes, training is done on different combinations of hyperparameters and the best performing model is found to be Model # 11 on the Excel sheet with the following key parameters and performance metrics:\n", "\n", "" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "LAHbHBnVEApj" }, "outputs": [], "source": [ "# Loading the data (5 Classes):\n", "model_number = 11 # See the excel sheet for list of other models\n", "X_train, X_test, X_val, y_train, y_test, y_val = data_loader(params_sheet, model_number)" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": {}, "colab_type": "code", "id": "YfsgsSgyEApj" }, "outputs": [], "source": [ "# Compiling the model (5 Classes):\n", "model, num_epochs, mini_batch_size, input_pixel = model_compiler(params_sheet, model_number)" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 1000 }, "colab_type": "code", "executionInfo": { "elapsed": 2506, "status": "ok", "timestamp": 1560171734300, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "MS6ZmZNuwmrB", "outputId": "3cb7ef0c-b744-4762-9fd6-b15aaf0251bc" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Training accuracy:\t(71.24%, 71.24%) (last epoch, highest)\n", "Validation accuracy:\t(75.69%, 75.96%) (last epoch, highest)\n", "Test accuracy:\t\t(75.66%)\n", "\n", "Model accuracy and loss function:\n", "\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0MAAAEWCAYAAACkHEyCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xd8ZHW9//HXJ5NkJpPeN22T7b0v\nRVDqIh1EkaaiCPLTa1fuFb1ewHYvevWqF1AvKiqoIIgooktTqoC07Y3tu9kkm2z6pifz/f1xJtns\nkt1Nm0zK+/l4nMc5M3PKZ2bLmc98v9/P15xziIiIiIiITDQx0Q5AREREREQkGpQMiYiIiIjIhKRk\nSEREREREJiQlQyIiIiIiMiEpGRIRERERkQlJyZCIiIiIiExISoZEejGzEjNzZhbbj30/YmYvjkRc\nIiIiQ6V7nMjbKRmSMcvMdplZu5llHfH8qvB/9iXRieywWJLM7KCZrYx2LCIiMnaM5nvcQJIqkdFO\nyZCMdTuBq7sfmNkCIBi9cN7mfUAbcI6ZTRrJC+smJSIy5o32e5zImKdkSMa6+4Brez3+MHBv7x3M\nLNXM7jWzKjPbbWZfNbOY8Gs+M/uumR0wsx3AhX0c+3MzKzezfWb2TTPzDSC+DwM/AdYCHzzi3EVm\n9odwXNVmdmev1z5mZpvMrNHMNprZ0vDzzsym99rvl2b2zfD2GWZWamZfMrMK4Bdmlm5mj4WvURve\nLux1fIaZ/cLMysKv/zH8/Hozu7jXfnHhz2jJAN67iIgMzWi/x72NmfnN7Afh+0pZeNsffi0rfB+q\nM7MaM3uhV6xfCsfQaGZbzOzsocQh0l9KhmSsewVIMbM54f/ArwJ+fcQ+dwCpwFTgdLwby3Xh1z4G\nXAQsAZYDlx9x7C+BTmB6eJ93Azf0JzAzKwbOAH4TXq7t9ZoPeAzYDZQABcAD4dfeD9wW3j8FuASo\n7s81gUlABlAM3Ij3b/wX4ceTgRbgzl7734f3K+M8IAf4fvj5ezk8ebsAKHfOrepnHCIiMnSj9h53\nDP8OnAwsBhYBJwJfDb/2RaAUyAZyga8AzsxmAZ8CTnDOJQPnAruGGIdIvygZkvGg+5ezc4BNwL7u\nF3rdPL7snGt0zu0Cvgd8KLzLFcAPnHN7nXM1wH/1OjYXLwn4nHOuyTlXiZcsXNXPuD4ErHXObcRL\ndOb1alk5EcgH/jV87lbnXPdA1RuA7zjnXnOebc653f28Zgi41TnX5pxrcc5VO+ceds41O+cagW/h\n3SwxszzgfODjzrla51yHc+658Hl+DVxgZim93st9/YxBRESGz2i9xx3NB4CvO+cqnXNVwNd6xdMB\n5AHF4XvOC845B3QBfmCumcU553Y557YPMQ6RftGYAhkP7gOeB6ZwRPcBIAuIw2uB6bYbryUGvIRk\n7xGvdSsOH1tuZt3PxRyx/7FcC/wUwDm3z8yew+visAooAnY75zr7OK4IGOxNoMo519r9wMyCeDe3\n84D08NPJ4RtoEVDjnKs98iTOuTIz+wfwPjN7BC9p+uwgYxIRkcEbrfe4o8nvI5788PZ/4/V8eDJ8\nzbudc7c757aZ2efCr80zsyeALzjnyoYYi8hxqWVIxrxwq8lOvF+4/nDEywfwfokq7vXcZA79slaO\nlxT0fq3bXrziB1nOubTwkuKcm3e8mMzsFGAG8GUzqwiP4TkJuCZc2GAvMPkoRQ72AtOOcupmDh88\ne2RRBnfE4y8Cs4CTnHMpwGndIYavk2FmaUe51q/wusq9H3jZObfvKPuJiEiEjMZ73HGU9RFPWfi9\nNDrnvuicm4rXBfwL3WODnHO/dc69M3ysA749xDhE+kXJkIwX1wNnOeeaej/pnOsCHgS+ZWbJ4XE8\nX+BQn+sHgc+YWaGZpQM39zq2HHgS+J6ZpZhZjJlNM7PT+xHPh4GngLl4/aYXA/OBBLxWllfxblK3\nm1mimQXM7NTwsT8DbjKzZeaZHo4bYDVeQuUzs/MId3k7hmS8cUJ1ZpYB3HrE+1sJ/ChcaCHOzE7r\ndewfgaV4LUJH/hopIiIjZ7Td47r5w/ev7iUGuB/4qpllm1cW/JbueMzsovA9zYB6vO5xITObZWZn\nhQsttOLdt0ID/IxEBkXJkIwLzrntzrnXj/Lyp4EmYAfwIvBb4J7waz8FngDWAG/y9l/drgXigY1A\nLfB7vP7OR2VmAbx+2nc45yp6LTvxujt8OHwDuxhv0OoevAGlV4bfy0N4Y3t+CzTiJSUZ4dN/Nnxc\nHV6/7D8eKxbgB3gJ2AG8gbiPH/H6h/B+VdwMVAKf637BOdcCPIzXNePIz0VEREbIaLrHHeEgXuLS\nvZwFfBN4Ha+K6rrwdb8Z3n8G8HT4uJeBHznnnsEbL3Q73r2qAq+gz5cHEIfIoJk3bk1E5O3M7BZg\npnPug8fdWURERGSMUQEFEelTuFvd9RyqAiQiIiIyrqibnIi8jZl9DG9w7Urn3PPRjkdEREQkEtRN\nTkREREREJiS1DImIiIiIyIQ05sYMZWVluZKSkmiHISIyob3xxhsHnHPZ0Y5jNNJ9SkQk+vp7nxpz\nyVBJSQmvv3606pIiIjISzGz38feamHSfEhGJvv7ep9RNTkREREREJiQlQyIiIiIiMiEpGRIRERER\nkQlpzI0ZEhERERGRvnV0dFBaWkpra2u0QxkRgUCAwsJC4uLiBnW8kiERERERkXGitLSU5ORkSkpK\nMLNohxNRzjmqq6spLS1lypQpgzqHusmJiIiIiIwTra2tZGZmjvtECMDMyMzMHFIrmJIhEREREZFx\nZCIkQt2G+l7VTU5EZDg4B40VcGALdLSCC3kL7tC2C2+HuiDUAV0dEOoMr3s9diGIiYUYX3gdd/jj\nuATImQPZs8E3uD7SMjq9uaeWJzZU8G/nzsYXM3G+zIiIRIuSIZHRJhSCjiZobYC2RmhrgFg/ZEwF\nf/IIxtEFdXugaov3Bd85mHU+ZM8a3Pk627xzxPphOH6xCoWgq807b1e7tw51QnwSBFK860RKezNU\nbYL9Gw5fWmoid82++PyQOw/yF0PeYm+dPQdi49++b0crtNYfWmJ8ULB0ZOOV49pU3sD/PbeDj5xS\nQl5qQrTDEREZsOrqas4++2wAKioq8Pl8ZGdnA/Dqq68SH9/HPeoI1113HTfffDOzZg3yO8cAKBkS\niaSOVqjeBs0HoKUWmmu8L8wtdeHtWu9x78SnrRFwfZ8vKRcypkFmeMmYBpnTIXkSdLSEz9EI7eF1\n28Hw44NeC0JsgpckxIXX3Y9jA3CwAqregqrNXvJzYCt0HtEH9+lbIXMGzLkY5lwE+UuPnth0tMDe\nV2Hn896y7w1wXYCFrx84fB2X4L0W6m4t6Xx760lX+6El1Hnsz97n95KiQCr4U7xtf4r3Wle79946\n23otrYfOaz6wGIiJ6bUdXne0QO2uQ39GcUHImet9HrnzvdYaf7L3uVjMoYXuxxZu5Yn1/kxi4sAX\nbv3pfmwWbkHqPPT+u1uTQp3en+v+9VC2CsrXwLrfw+v3hN93vNdqFBN3ePLT1Xb451N4Itzw1LE/\nQxlxhelBAPbWtCgZEpExKTMzk9WrVwNw2223kZSUxE033XTYPs45nHPExPQ9YucXv/hFxOPspmRI\npLf2Zqjc6H3BrFgLFeu8L7CZMyBreng9A9KnQFzg8GNb6739y9d6x5av9ZKKvr60xyVCQjoE0711\n1nTwp3pfov3J4S/uyd6Xd3+Kl8zUbIfqHd76rcehqSoyn0HqZK/1Z8rp3jprFmTP9JKAzX+BTX+G\nf/wQXvwfSCmA2RfC7Iug6ETvPe98HnY+5yVCXW1eMlGwDE79jPeeOlqhs8U7X0crdDR7iUhHs9dy\n5Es5lCD44t+eLPjivQTO5w8neH7vOV+8l2C0N0FbfTjBbPDWrfXedkO5l2h0Hx/rh2Cil5D54r11\nTIwXR6gr3LWtq9d2yLvmoqu8BCh3nvd34Sj/mQ+J+bwE7GgtXDmzYcHl3nYoBLU7oXw1lK0O/701\nSCvyksHDljRvnZQ7/DHLkBWlewlQaW0zJ07JiHI0IiLDZ9u2bVxyySUsWbKEVatW8dRTT/G1r32N\nN998k5aWFq688kpuueUWAN75zndy5513Mn/+fLKysvj4xz/OypUrCQaD/OlPfyInJ2fY4lIyJONP\n97gM57wvskf+8t/ZGl7avS/OVZsOJTAH3gqP88D7wjhpobe94xlY89tD17AYSC3yEqO4oPfls3bn\nodeTcr1jZ53nfWFOyoWEDAhmeF9Gj0ykBqO1Hmp2QPV2OFgJ8YngT/KSp/ikcDKVBPHJ3muhzl7v\nvc1LRro/l84WCGZC1kxv374kpMOJH/OW5hp46wkvMXrzXnj17l47Gkxa4O035XQofsfIdu+biGJi\nDrUWzn9ftKORIchP85KhvTUtUY5ERMaDr/15AxvLGob1nHPzU7j14nmDOnbz5s3ce++9LF++HIDb\nb7+djIwMOjs7OfPMM7n88suZO3fuYcfU19dz+umnc/vtt/OFL3yBe+65h5tvvnnI76ObkiEZfdqb\nYdcLsP3v0FgebkFo8VoOutft4e2u9rcPVB+M5HzIWwhzLvHWkxZC2uTDu4C1NXpd3g5sg+qtXjey\n6q1eLHkLYckHIW+Rd2zyCPzqHkiF/CXe0l/+pOG5djADFl/tLe1NsO1vUPamF0vJu7zXRWTAAnE+\nclP8lNY2RzsUEZFhN23atJ5ECOD+++/n5z//OZ2dnZSVlbFx48a3JUMJCQmcf/75ACxbtowXXnhh\nWGNSMiSjQ/V22PoUbH0Sdr3oda+KC3qtL3EJ3nZCOqTke9vdz/niwmM6rI+xGeHxGbGB8LgY/+Hd\noWLjvTEzmdMhKfv4MfqTB558TATxiTD3Em8RkSErTA+yV8mQiAyDwbbgREpi4qHeJ1u3buWHP/wh\nr776KmlpaXzwgx/sc76g3gUXfD4fnZ3HGTM8QEqGZGCcg+ZqaCjzBv93NHstAz2tNU3hdbM38Ls7\naelex/faDnV5LUBbn/S6e4GXmJxwPcw4ByafMjzdyURExpDC9ATe2F0b7TBERCKqoaGB5ORkUlJS\nKC8v54knnuC8884b8TiUDMnbtTfDjme9BKWhDBrLvIHnjWXePCpd7cc5gXljVnyxhwbLH01swOtW\nddInYMYKr3y0iMgEVpQe5LG15XR2hYj1aW50ERmfli5dyty5c5k9ezbFxcWceuqpUYlDyZB4Olq8\nbmobHvEqlXWEu2jEJkBKnjempugkSM7zuqol53kD7uODXmW03uvYwOFjbUKhXtXDeo376eqESfPD\nJZVFRAS8lqGukKO8vpWijGC0wxERGbTbbrutZ3v69Ok9JbcBzIz77ruvz+NefPHFnu26urqe7auu\nuoqrrrpqWGNUMjSRdbTC9r95CdCWlV755mAmLLwS5r3HKwYQSBv6BJkxMd64kqNVKRMRkR7dCVBp\nbYuSIRGRCFMyNB5tWQnr/9CrqEC4wECM79AEki213lidtgavMMH898G8y7wuaz79tRARiZbC8FxD\ne2ubeQeZUY5GRGR807fe8cQ5ePlOePKrkJgdnlsmdGjiSBc6NHmkL96r/jXvMm8uGF9ctKMXEREg\nLzWBGPNahkREJLKUDI0XoS5Y+SV47acw9z1w2f+pEpuIyBgUHxvDpJQApTUqry0iEmlKhsaD9ib4\n/fXw1ko45TOw4mveOB0RERmTCjOCahkSERkBSobGusb9cP+VUL4GLvgunPixaEckIiJDVJiewMvb\nq6MdhojIuKfmg7Gsagv8fIW3vuq3SoRERMaJovQgFQ2ttHeGoh2KiMiAVFdXs3jxYhYvXsykSZMo\nKCjoedzefry5Kg+55557qKioiGCkHrUMjVW7XoQHrgGfHz7yFyhYGu2IRERkmBSmJ+AclNW1UJKl\naQlEZOzIzMzsmU/otttuIykpiZtuumnA57nnnntYunQpkyZNGu4QD6NkaDRqroHWuvDkpEdMVNre\nDA1l8Px3IL0EPvB7SC+OdsQiIjKMes81pGRIRMaLX/3qV9x11120t7dzyimncOeddxIKhbjuuutY\nvXo1zjluvPFGcnNzWb16NVdeeSUJCQm8+uqrxMfHRyQmJUOjRUMZbPyTNwHq3n8ef/+Sd8GV93lz\nBImIyLjSPddQaa0qyonIEKy8GSrWDe85Jy2A828f8GHr16/nkUce4aWXXiI2NpYbb7yRBx54gGnT\npnHgwAHWrfPirKurIy0tjTvuuIM777yTxYsXD2/8R1AyFE2NFbDxUdjwB9jzsvdc7gI4898htQji\nEiAuCPHBQ9vd68RsbyJVEREZdyalBPDFGHuVDInIOPH000/z2muvsXz5cgBaWlooKiri3HPPZcuW\nLXzmM5/hwgsv5N3vfveIxqVkaKS11MH638P6R2D3PwAHOXO9BGjueyB7ZrQjFBGRKIv1xZCfFlB5\nbREZmkG04ESKc46PfvSjfOMb33jba2vXrmXlypXcddddPPzww9x9990jFpeSoZFSvR1e+TGs/i10\nNEHWTDj9SzDvMsiZHe3oRGScc87R2NZJZUMblQ2txMQYJ0/NjHZYcgyFaUH2auJVERknVqxYweWX\nX85nP/tZsrKyqK6upqmpiYSEBAKBAO9///uZMWMGN9xwAwDJyck0NjZGPC4lQ5HkHOx8Hl75Ebz1\nBPjiYP7lcNL/g7xF6uYmIsOivTNEZWMr5fXeUlHfwv6GNvY3tFLZ6CU/+xvaaOno6jlmcVEaf/zk\nqVGMWo6nKCOBZ7dURTsMEZFhsWDBAm699VZWrFhBKBQiLi6On/zkJ/h8Pq6//nqcc5gZ3/72twG4\n7rrruOGGG1RAYUzqaPW6wr3yY9i/HoJZXivQ8o9Ccm60oxORMaSjK0RFfXei0+Kt67x1RYP3/IGD\nbTh3+HEJcT5yU/zkpARYUJjGimQ/OSl+clMC5CQHyE8LROcNSb8VpgepbGyjtaOLQJwv2uGIiAzY\nbbfddtjja665hmuuueZt+61ateptz11xxRVcccUVkQqth5Kh4bbqN/D0rdBUBTnz4NK7vNagOH3x\nEJG3a2ztoLS2Jbw0s6+2hX11LZSFk56qPhKdZH8seWkBJqUmMGdSCnlpAfJSvcd5qQFyUwKkBGIx\ntT6PaUUZXkW5fXUtTMtOinI0IiLjU0STITM7D/gh4AN+5py7/YjXvw+cGX4YBHKcc2mRjCmitqyE\nP30SJr8D3vdzmHKausKJTEChkKO+pYPqpnaqD7ZR09ROdVO7tz7YRnl9K/vqvASovqXjsGMT4nzk\npwXIT0tg1qxs8lITyA8nPvmpAfLSEkjy63esiaAw3ZtraG9Ns5IhEZEIidgd1cx8wF3AOUAp8JqZ\nPeqc29i9j3Pu8732/zSwJFLxRFzFenj4BshfDB982CuHLSLjVmtHF7urm9l5oImdB5rY1b2ubqK6\nqZ2ukOvzuORALHmpAQrSElg6OZ3C9AQK04PhdQIZifFq0REAitIPTbwqIjIQ3eNvJgJ3ZPeJAYrk\nz4snAtucczsAzOwB4FJg41H2vxq4NYLxRE7jfrj/KvCnwFX3KxESGSc6u0KU1raws7qJnVVNPYnP\nzgNNlNW3HNZ9LSvJz5SsIKfPzCY3JUBGYjyZSfFkJvp7ttOD8cTHxkTvDcmYkpPsJ94Xo7mGRGRA\nAoEA1dXVZGZmjvuEyDlHdXU1gcDgh6NEMhkqAPb2elwKnNTXjmZWDEwB/n6U128EbgSYPHny8EY5\nVB0t8MA10FwN162ElLxoRyQiA9Da0UVpbQt7a5sprW1hT7WX7Ow40MTemmY6ug5lPMmBWKZmJXJC\nSTolWYVMyUpkalYSJVlBkgNxUXwXMh7FxBgF6QlqGRKRASksLKS0tJSqqolRjTIQCFBYWDjo40dL\nx/OrgN8757r6etE5dzdwN8Dy5cuH1hY2nJyDP30K9r0OV/7a6yInIqNSa0cXT27cz6byBi/5qfGS\nnwMH2w7bzx8bw5SsRGblJnPevEmUZCUyNSuRKVmJ6sI2zpjZPcBFQKVzbv4x9jsBeBm4yjn3+5GK\nD6BQyZCIDFBcXBxTpkyJdhhjRiSToX1AUa/HheHn+nIV8MkIxhIZz33HK6F99q0w5+JoRyMifdhb\n08xv/rmHB1/fS01TO7HhX9sL0xNYMSenZ8xOUYa3zk7yExOjhGeC+CVwJ3Dv0XYIj3/9NvDkCMV0\nmML0IE9uqIjGpUVEJoRIJkOvATPMbApeEnQV8LbC4mY2G0jH+9Vt7Fj/B3j2P2HR1fDOzx9/fxEZ\nMaGQ44VtB7jv5V38bXMlBpwzN5dr31HCyVMz8SnZEcA597yZlRxnt08DDwMnRDygPhSmJ1Dd1E5z\neyfB+NHSmUNEZPyI2P+szrlOM/sU8AReae17nHMbzOzrwOvOuUfDu14FPOCGWgpiJJW+AX/8hFdC\n++Ifqny2yChR39zBQ2/s5Tf/3MPOA01kJcXzyTOmc81Jk8lPS4h2eDLGmFkBcBneFBDHTIYiNba1\nKONQRbmZucnDdl4REfFE9Gcm59xfgb8e8dwtRzy+LZIxDLv6Uq9yXFKuN04o1h/tiETGtd5z9tQ0\ntXPgYBtVjW19rNupbGylo8uxrDidz62YwXnzJ+GP9UX7LcjY9QPgS8650PHGikVqbGthupfEl9Y2\nKxkSEYkAtbkPRCgED3wAOlvhw49CYla0IxIZs5xz1DS1s7ummT3VzeyqbmJ/Qxu14aSnptlb1zW3\n09eUPTEGmUl+spL8ZCf7mZaTxKSUABcuzGNefurIvyEZj5YDD4QToSzgAjPrdM79caQC6E6G9tao\niIKISCQoGRqIvf+E8tVwyR2QMyfa0YiMWm2dXdQ1d1Db3E5tUwd1zV5y45Wu9hKfPdXNNLZ19hxj\nBpmJ8WQkevPxzMhJIqPX4+55erKTveQnPRivsT8SUc65nnJMZvZL4LGRTIQAspP8+GNjKNVcQyIi\nEaFkaCDW/g7igjDvvdGORCTqGlo72FjWwIayBjbsq2dr5UFqmtqpbW6nub3PKvnE+YzC9CDFmUGW\nF6czOTORkkzvcWF6kECcurTJyDGz+4EzgCwzK8Wb+DsOwDn3kyiG1sPMKExPUMuQiEiEKBnqr842\n2PAIzL4I/EnRjkZkSJraOtlT08zu6mb21jSzp6aZLudI8seS5I8l0R9Lkt9Hkj+ORL+PJH8sTe1d\nbCirZ8O+BtaX1bO7+tAv1bkpfmZNSmFmbjJpwTjSg3GkBb2WnJ7txDiyk/zE+mKi+M5FDnHOXT2A\nfT8SwVCOqSgjSGmdWoZERCJByVB/bX0KWutg4ZXRjkTkMKGQ42B7JwdbO2ls7eRgWwcNrYc/rm/p\nYF9tC7trvOTnwMH2w86REoglPjaGg22dtHaEjnm9oowE5uencsXyIubmpzAvP4Wc5EAk36LIhFaY\nnsCqPXXRDkNEZFxSMtRfa38Hidkw9YxoRyITQENrB3t6tdpUNbZR39JBQ2sHDS2dNLR6CU5DSweN\nbZ0crzB9jEF+WgLFmUHOmZtLUUaQ4oxEJmcEmZwRJDUY17NvZ1eIpraungTrYJu3xPtimJuXcti+\nIhJ5RenBnn//KQH9+xMRGU5KhvqjpQ7eehyWXw8+fWQyeEeWia4+2MaBpnbK6lrYU3Mo+alr7jjs\nuGC8j9SEOFICcaQkxDIpJcDM3GRSArGkJsSRHIgjORBLUiC2ZzvZ720nBWJJjPdxvNLA3WJ9MaQG\nY5T0iIwSheneXEP7altIydO/SxGR4aRv9v2x8U/Q1Q4Lr4h2JDJKHWzrpLKhlcrGNvY3tFLV2HbY\ndvXBdqrDxQW6+qgT3V1YoCgjyMLC1J4Wm6Lwol+DRSauoozu8trNzMlLiXI0IiLji5Kh/lj7IGTO\ngPwl0Y5EouRgW2dPq83emmavRHR4u6yuhaY+qqfFx8aQm+InO8nP5MwgS4vTyEiMJzPRT2ZS/GHb\nWUl+lYkWkT51twyV1qqinIjIcFMydDx1e2H3i3DmV72JUGTc6OwKUVrbQmVjmzfJZ1M7NU1tPV3Y\nvG5s7ZTXt1B7RLe1JH8sRRlBpmQl8s4ZWeSmBMhJ9pOTHCA3xVunJMT2u2uaiMjRpAfjSIz3sVdz\nDYmIDDslQ8ez7iFvveDy6MYhg9bc3smOqia2VR5kW+VBtld5613VTXR0vb3LWpI/tmeyz0mpARYV\npYW7rCV46/QgacE4JToiMiK8uYaCahkSEYkAJUPH4pxXRa7oZMiYcvz9ZcS1dXZR2dBGRUMrFfWt\n7A+vKxq87bK6VvbVHfoC4YsxijOCTM1O4uw5uUzNTmRSSsDrspbkzYujiT9FZLQpykhgb41ahkRE\nhpuSoWOpWAdVm+HC/4l2JBOec479DW3epJ9lDazf5617JzrdAnExTEoJkJsS4ISSdK7KLmJ6ThLT\ncpIozgzij1WyIyJjS2F6kH/uqME5p1ZpEZFhpGToWNb+DmLiYN5l0Y5kQmls7WDXgWZ2VjexqbyB\nDWUNbCyr75ko1AymZCayZHIaVywvIi81QG5qgEkp3qKxOiIy3hSmJ9DY1kl9SwdpwfhohyMiMm4o\nGTqaUBes+z3MeDcEM6IdzbjT3hnirf2N7DzQxK4DTeyqbmZXtbdd3dTes19sjDEjN5kzZ+UwLz+F\neQWpzMlLIcmvv7oiMnH0riinZEhEZPjoG+XR7HweDlZobqFhUl7fwqo9dazaU8uqPXWs21dPW2eo\n5/VJKQGKM4OcMzeXkqxESjKDlGQlMiUrUd3aRGTCK0w/NNfQ/ILUKEcjIjJ+KBk6mrUPgj8FZp4X\n7UjGnINtnWzYV8/a0nreDCc/FQ2tgDf3zvz8FD50cjGLitKYHh7HE4zXX0URkaMpytBcQyIikaBv\noH1pb4ZNj3pjheIC0Y5mVGtq62RDWQNrS+tYv6+etfvq2XmgCReuWF2UkcCJUzJYMjmNJZPTmZOX\nrJYeEZEBSk2IIzkQq7mGRESGmZKhvmz5K7QfhIVXRjuSUcU5x/aqg7y+q5bXdtWyprSO7VUHexKf\nSSkB5hek8p7FBSwoSGV+QSrZyf7oBi0iMk4Uaa4hEZFhp2SoL2sfhJQCKD412pFEVVtnF+v31fck\nP2/srqG2uQOAjMR4lhSlcdFGIsaxAAAgAElEQVTCPBYWeolPTrJa0UREIqUwPYFd1U3RDkNEZFxR\nMnSkpgOw7Wk45dMQExPtaEZUeX0Lq/fUsWqvV+hgTWk97eEiB1OyElkxJ5flJeksL8lgalaiyleL\niIygoowgL2w9oLmGRESGkZKhI63/A7iucV9Frrm9k3Wl9azaWxdOgGrZ39AGQLwvhrn5KVx7cjHL\nS9JZVpyh7m4iIlFWmJ5AS0cXNU3tZCbp/2QRkeGgZOhIa38HufMhd160IxlW9c0dvLarhn/urObV\nnTWsL2ugK+QN9pmcEeTkqZksLlKRAxGR0aooPNfQ3toWJUMiIsNEyVBv1dth3+twztejHcmQHTjY\nxqs7a3h1Zw3/3FnD5ooGnPNafRYXpfHx06eydHI6i4vSdFMVERkDCjO8uYZKa5tZXJQW5WhERMYH\nJUO97f6Ht551QXTjGITOrhCr9tbx982VPLO5ks0VjQAkxPlYVpzO51fM5MQpGSwuSiMQp1YfEZGx\nprC7ZahGFeVERIaLkqHeylZ5E61mTIt2JP1S29TOc29V8ffNlTz3VhX1LR34Yozlxen823mzOHlq\nJvPzU4mPnViFIERExqMkfyzpwThKNdeQiMiwUTLUW9kqyFs0qqvIhUKO37y6hz+u2seqPbWEHGQm\nxnP2nBzOmp3Du2Zkk5oQF+0wRUQkAooyguzVXEMiIsNGyVC3zjaoWA8nfyLakRxVeX0LX3xwDS9t\nr2ZuXgqfOmsGZ83OYWFBKjExKrMqIjLeFaYn9HSDFhGRoVMy1K1yI4Q6IH9JtCPp08p15dz8h3W0\nd4b49vsWcMXyIs0zISIywRSmB3l6UyWhkNOPYCIiw0DJULeyVd56lCVDTW2dfP3PG/nd63tZWJjK\nD65czNTspGiHJSIiUVCUnkB7Z4gDB9vISQlEOxwRkTFPyVC3slWQkA7pJdGOpMeavXV89oFV7K5p\n5pNnTuNzK2YS5xu945lERCSyCnvNNaRkSERk6JQMdStb5bUKjYKuZ10hx0+e2873n3qLnGQ/93/s\nZE6emhntsEREJMomZ3rJ0Fv7G1lWnB7laERExj41MwB0tEDlplHRRW7XgSauvvsV/vuJLZw3fxIr\nP3uaEiEREQFgalYixZlB/rquPNqhiIiMC2oZAti/AUKdUU2GOrpC/OyFnfzg6beI98Xwvfcv4r1L\nC1QkQUREepgZFy/M50fPbqOqsY3sZH+0QxIRGdPUMgRRL56wrrSeS+/8B99+fDOnz8zmqS+czvuW\nFSoREhGRt7lkcT4hh1qHRESGgVqGAPa9CYnZkFIwopdtae/i+0+/xc9e2EFmkp+ffHAp583PG9EY\nRERkbJmZm8zsSck8uqaMD59SEu1wRETGNCVDEJXiCS9sreIrj6xjb00LV59YxM3nzyE1IW7Eri8i\nImPXxYvy+e8ntlBa29xTYU5ERAZO3eTaDsKBLSPWRa6uuZ0vPriGD/38VWJjYnjgxpP5r/cuVCIk\nIiL9dsmifAD+vEZd5UREhiKiyZCZnWdmW8xsm5ndfJR9rjCzjWa2wcx+G8l4+lSxDlxoRJKhV3ZU\nc/4PX+BPq/fxyTOnsfKz71KlOBERGbCijCBLJqfx6JqyaIciIjKmRaybnJn5gLuAc4BS4DUze9Q5\nt7HXPjOALwOnOudqzSwnUvEc1QgUT+joCvGDp9/iR89upyQzkT/8yyksLEyL2PVERGT8u2RRPl/7\n80a2VTYyPSc52uGIiIxJkWwZOhHY5pzb4ZxrBx4ALj1in48BdznnagGcc5URjKdvZasgOR+SJ0Xk\n9Lurm7j8Jy9z1zPbef+yQh779DuVCImIyJBduDCPGINHV6t1SERksCKZDBUAe3s9Lg0/19tMYKaZ\n/cPMXjGz8/o6kZndaGavm9nrVVVVwxtld/GEYeac4+E3Srnghy+ws+ogd12zlO9cvohEv2pWiIjI\n0OUkB3jHtEweXVOGcy7a4YiIjEnRLqAQC8wAzgCuBn5qZm9rNnHO3e2cW+6cW56dnT18V29tgOqt\nw54MNbR28NkHVvPFh9YwryCVlZ87jQsXqmS2iMhoYmb3mFmlma0/yusfMLO1ZrbOzF4ys0UjHePx\nXLIon13VzazbVx/tUERExqRIJkP7gKJejwvDz/VWCjzqnOtwzu0E3sJLjkZG+RpvPYzJ0Jq9dZz/\ngxf4y7pybnr3TO7/2MkUpCUM2/lFRGTY/BLos0dC2E7gdOfcAuAbwN0jEdRAnDcvjzifqauciMgg\nRTIZeg2YYWZTzCweuAp49Ih9/ojXKoSZZeF1m9sRwZgO11M8YfGwnK6xtYMb73sdgIc+/g4+ddYM\nfDEjN3eRiIj0n3PueaDmGK+/1D2mFXgF70e9USU1GMfpM3N4bG05oZC6yomIDNRxkyEz+7SZpQ/0\nxM65TuBTwBPAJuBB59wGM/u6mV0S3u0JoNrMNgLPAP/qnKse6LUGrWwVpE6GxKxhOd33nnyLysY2\n7vrAUpZOHvBHJiIio9f1wMqjvRjRsa3HccnifCoaWnl111HzOhEROYr+tAzl4pXFfjA8b1C/mzqc\nc391zs10zk1zzn0r/NwtzrlHw9vOOfcF59xc59wC59wDg3sbg1T25rC1Cq0trePel3fxoZOLWVyk\nanEiIuOFmZ2Jlwx96Wj7DNvY1k1/hnvOg66Ofh+yYk4OCXE+zTkkIjIIx02GnHNfxRvH83PgI8BW\nM/tPM5sW4dgiq7kGancNy3ihrpDj3x9ZT2aSn5vOnTX02EREZFQws4XAz4BLR6TnQmcb7HkZKjf1\n+5BgfCwr5uaycl05HV2hCAYnIjL+9GvMkPNqdlaEl04gHfi9mX0ngrFFVvlqb12wdMinuu/lXazb\nV88tF80lJRA35POJiEj0mdlk4A/Ah5xzb43IRQuWeet9bwzosEsW5VPb3MGLWw9EICgRkfGrP2OG\nPmtmbwDfAf4BLHDOfQJYBrwvwvFFTnfxhLyhVUqtqG/lu0++xWkzs7lI5bNFRMYMM7sfeBmYZWal\nZna9mX3czD4e3uUWIBP4kZmtNrPXIx5UegkkZAw4GTptZhYpgVh1lRMRGaD+zACaAbzXObe795PO\nuZCZXRSZsEZA2SrImAoJQyt08PXHNtDRFeIbl85jAMOpREQkypxzVx/n9RuAG0YoHI+Z1zq0780B\nHeaP9XH+/DweW1tGS3sXCfG+CAUoIjK+9Keb3Ep6lR41sxQzOwnAOdf/Ts2jTdnqIY8XemZzJX9d\nV8Gnz5pOcWbiMAUmIiITWsEyqNoEbQcHdNgli/Npau/i75srIxSYiMj4059k6MdA7/+RD4afG7sO\nVkH93iElQy3tXfzHn9YzPSeJG08b27UkRERkFClYBi50aGLwfjp5aibZyX4eXXPk/OYiInI0/UmG\nLFxAAfC6x9G/7nWjV3fxhCEkQ3f8fSultS188z3ziY+N5Ny1IiIyoXQX9ikbWFc5X4xx4YI8ntlS\nRUNr/0tzi4hMZP35Fr/DzD5jZnHh5bPAjkgHFlFlqwCDSQsHdfhb+xu5+/kdXL6skJOnZg5vbCIi\nMrElZkHa5AEXUQCvq1x7Z4gn1ldEIDARkfGnP8nQx4FTgH1AKXAScGMkg4q4slWQNQMCKQM+NBRy\n/Psj60gKxPKVC+ZEIDgREZnwCpYNKhlaUpRGcWaQe1/eTa9OHSIichT9mXS10jl3lXMuxzmX65y7\nxjk3tkdnlq0adBe5379Rymu7avnK+XPISIwf5sBERETwkqG6Pd4Y1wEwMz591gzW7atnpVqHRESO\nqz/zDAXM7JNm9iMzu6d7GYngIqKhHBrLIX/gk622dXZx++ObOaEkncuXFUYgOBERGQwzm2Zm/vD2\nGeHu3WnRjmvQuidfHeC4IYDLlhQwIyeJ7z65hc6u0DAHJiIyvvSnm9x9wCTgXOA5oBBojGRQEdU9\n2eogWoZe2l5NTVM7nzhjGjExmlNIRGQUeRjoMrPpwN1AEfDb6IY0BHmLwGIG1VXOF2PcdO4sdlQ1\n8fCbpREITkRk/OhPMjTdOfcfQJNz7lfAhXjjhsamslXeDWbSggEf+vi6CpL8sZw6PSsCgYmIyBCE\nnHOdwGXAHc65fwXyohzT4MUnQs7cQSVDAO+em8viojR+8PRWWju6hjk4EZHxoz/JUHd9zjozmw+k\nAjmRCynCylZB9hyIDw7osM6uEE9t2s9Zs3Pwx2pmbxGRUabDzK4GPgw8Fn4uLorxDF3+Etj3Jgyi\nEIKZ8W/nzaK8vpVfv7I7AsGJiIwP/UmG7jazdOCrwKPARuDbEY0qUpwbdPGEV3fVUNPUznnzJ0Ug\nMBERGaLrgHcA33LO7TSzKXjdvMeugmXQUgO1uwZ1+CnTsnjXjCzuemYbjZp3SESkT8dMhswsBmhw\nztU65553zk0NV5X7vxGKb3jVl0LzAchfPOBDn1hfgT82hjNmZUcgMBERGQrn3Ebn3Gecc/eHf8BL\nds6NzR/uunUXURhkVzmAfz13FrXNHfz0hZ3DFJSIyPhyzGTIORcC/m2EYom8nuIJA6skFwo5Ht9Q\nwekzswnGx0YgMBERGQoze9bMUswsA3gT+KmZ/U+04xqSnDkQm+B1lRukhYVpXLBgEj97YQcHDrYN\nY3AiIuNDf7rJPW1mN5lZkZlldC8RjywSylZBTCzkzhvQYatL69jf0Mb5C9RFTkRklEp1zjUA7wXu\ndc6dBKyIckxD44vzqsoNoWUI4AvnzKK1o4u7ntk2TIGJiIwf/UmGrgQ+CTwPvBFeXo9kUBEz7zK4\n+IcQFxjQYU+sryA2xjhrdm6EAhMRkSGKNbM84AoOFVAY+wqWQfka6Br8mJ/pOUm8f1kRv3llD6W1\nzcMYnIjI2HfcZMg5N6WPZepIBDfs8hbCkg8O6BDnHCvXV3DK9CxSE8Z2YSIRkXHs68ATwHbn3Gtm\nNhXYGuWYhq5gKXS2QOWmIZ3msytmgMEPnh77H4mIyHA67gAYM7u2r+edc/cOfzijz6byRvbUNPOJ\nM6ZFOxQRETkK59xDwEO9Hu8A3he9iIZJQXiM6743vB/0Bik/LYFrTy7mnn/s5P+dNpUZucnDFKCI\nyNjWn25yJ/Ra3gXcBlwSwZhGlcfXlxNjcM5cdZETERmtzKzQzB4xs8rw8rCZFUY7riFLnwIJ6VA2\n+CIK3f7lzOkE42P57pNbhiEwEZHxoT/d5D7da/kYsBRIinxoo8PjGyo4oSSDrCR/tEMREZGj+wXe\nXHj54eXP4efGNjNv3NAQKsp1y0iM52PvmsoTG/azak/tMAQnIjL29adl6EhNwJThDmQ02l51kLf2\nH9REqyIio1+2c+4XzrnO8PJLYHxMDFewDCo3QnvTkE91/bumkJkYz3/+dRMdXaFhCE5EZGw7bjJk\nZn82s0fDy2PAFuCRyIcWfY+vrwDg3HlKhkRERrlqM/ugmfnCyweB6mgHNSwKloELeVXlhijJH8uX\nL5jDa7tq+eKDa+gKuWEIUERk7OrPDKLf7bXdCex2zpVGKJ5R5fH1FSwqSiM/LSHaoYiIyLF9FLgD\n+D7ggJeAj0QzoGGT36uIQvEpQz7d5csKqWxs5TuPbyE5EMs33zMfMxvyeUVExqL+JEN7gHLnXCuA\nmSWYWYlzbldEI4uy0tpm1u2r5+bzZ0c7FBEROQ7n3G6OKO5jZp8DfhCdiIZRUjakTh7y5Ku9/csZ\n02lo6eQnz20nJSGOL52ne52ITEz9GTP0ENC7Y3EXvcqXjldPbNgPqIuciMgY9oVoBzBsCpYOazIE\n8KXzZvGBkybz42e386Nntw3ruUVExor+JEOxzrn27gfh7fjIhTQ6PL6+nNmTkpmSlRjtUEREZHDG\nT9+vgmVQtweaDgzbKc2Mb1w6n0sX5/Odx7dw3yu7h+3cIiJjRX+SoSoz6+l6YGaXAsP3v/EoVNnY\nyuu7a1VFTkRkbBs/1QEKlnnrYSix3VtMjPHd9y/i7Nk53PKn9fxx1b5hPb+IyGjXn2To48BXzGyP\nme0BvgT8v8iGFV1PbtiPcygZEhEZ5cys0cwa+lga8eYbGh/yFoHFDHtXOYA4Xwx3fWApJ03J4IsP\nreHpjfuH/RoiIqNVfyZd3e6cOxmYC8x1zp3inBvXnYuf2FDBlKxEZuUmRzsUERE5BudcsnMupY8l\n2TnXnyJBY4M/CbLnRCQZAgjE+fjZh09gfn4K//LbN3lp27juACIi0qM/8wz9p5mlOecOOucOmlm6\nmX1zJIKLhrrmdl7eXs258yap1KiIiIwe3UUUXGR6/yX5Y/nldSdSkhnkhntf58kNFRG5jojIaNKf\nbnLnO+fquh8452qBCyIXUnQ9vamSzpDjfHWRExGR0aRgKbTUQO2uiF0iPTGeX19/EtOyk7jxvjf4\n9uOb6ewKHf9AEZExqj/JkM/M/N0PzCwB8B9j/zHt8fUV5KcGWFiYGu1QREREDukpohCZrnLdclIC\nPPTxd3D1iUX8+NntXHvPqxw42BbRa4qIREt/kqHfAH8zs+vN7AbgKeBXkQ0rOg62dfL81irOna8u\nciIiMsrkzIXYwLBXlOtLIM7Hf713Id+5fCFv7K7lov99kTf31Eb8uiIiI60/BRS+DXwTmAPMAp4A\niiMcV1Ss2lNLe2eIs2fnRjsUERGRw/nivKpyZZFPhrpdsbyIhz9xCnGxxpX/9zK/emkXLkJjlkRE\noqE/LUMA+/Hma3g/cBawKWIRRVF5fSsAxZnBKEciIiLSh4JlULYaujpH7JLzC1J57FPv4l0zsrn1\n0Q18/neraW4fueuLiETSUZMhM5tpZrea2WbgDmAPYM65M51zd/bn5GZ2npltMbNtZnZzH69/xMyq\nzGx1eLlh0O9kGFQ2eMlQdvK4HRIlIiJjWcEy6GyBqpH9TTI1GMfPrl3OF8+ZyZ/WlHHZXS+xpaJx\nRGMQEYmEY7UMbcZrBbrIOfdO59wdQFd/T2xmPuAu4Hy8OYquNrO5fez6O+fc4vDyswHEPuwqGlpJ\nC8YRiPNFMwwREZG+dRdR2P3yiF86Jsb49Nkz+NV1J1LZ2MqF//sC3358My3t/f5qICIy6hwrGXov\nUA48Y2Y/NbOzgYFUFTgR2Oac2+GcawceAC4dfKiRt7+hjUkpgWiHISIi0reMKZA+Bbb/LWohnDYz\nm6e/cDrvWVLAj5/dzjnff46/b94ftXhERIbiqMmQc+6PzrmrgNnAM8DngBwz+7GZvbsf5y4A9vZ6\nXBp+7kjvM7O1ZvZ7Myvq60RmdqOZvW5mr1dVVfXj0oNT2dBKjpIhEREZzaavgJ3PQ0dr1ELITPLz\n3fcv4oEbTyYQ5+Ojv3ydj9/3BuX1LVGLSURkMPpTTa7JOfdb59zFQCGwCvjSMF3/z0CJc24hxyjZ\n7Zy72zm33Dm3PDs7e5gu/XYVDa3karyQiIiMZjPOgY5m2PNStCPh5KmZ/PUz7+Jfz53FM1sqWfG9\n5/j5izs1UauIjBn9rSYHgHOuNpyYnN2P3fcBvVt6CsPP9T5ftXOueya3nwHLBhLPcOoKOaoa28hV\ny5CIyIRgZveYWaWZrT/K62Zm/xsuArTWzJaOdIx9Knkn+PywLXpd5XqLj43hk2dO56nPn87ykgy+\n8dhGLrnzH/xzR7XKcIvIqDegZGiAXgNmmNkUM4sHrgIe7b2DmeX1engJUSzZXX2wjZCD3FQlQyIi\nE8QvgfOO8fr5wIzwciPw4xGI6fjiE6H4FNj6VLQjOczkzCC/vO4EfvSBpVQ3tXHl3a9w8Z0v8vs3\nSmntUJEFERmdIpYMOec6gU/hTdK6CXjQObfBzL5uZpeEd/uMmW0wszXAZ4CPRCqe46kIl9VWNzkR\nkYnBOfc8UHOMXS4F7nWeV4C0I37Ei54Z58CBLVC3J9qRHMbMuGBBHs/cdAbfumw+bR0hbnpoDafe\n/ne+9+QW9jdEb5yTiEhfItkyhHPur865mc65ac65b4Wfu8U592h4+8vOuXnOuUXh+Ys2RzKeY9nf\n4PXWUzc5EREJ628hoBEr9NNj+gpvve3pyF9rEILxsXzgpGKe/Pxp/OaGk1gyOY07n9nGqbf/nc/c\nv4o399SqC52IjAqx0Q5gtOj+tWqSusmJiMgAOefuBu4GWL58eeS/5WfNhNTJsPVpWP7RiF9usMyM\nU6dncer0LHZXN3Hvy7t58LW9PLqmjIWFqVxz4mQuXpRPol9fR0QkOiLaMjSWVDa0EmOQmRgf7VBE\nRGR0OG4hoKgxgxkrYOdz0Nke7Wj6pTgzkf+4aC6vfOVsvn7pPFo7urj5D+s48VtP8+U/rGNdaX20\nQxSRCUjJUFhFQytZSX5iffpIREQE8Ir+XBuuKncyUO+cK492UD2mr4D2g7D3lWhHMiCJ/liufUcJ\nT3zuNB7+xDs4f0Eej6wq5eI7X+TC/32BX7+ym8bWjmiHKSIThNqlw/Y3tKmLnIjIBGJm9wNnAFlm\nVgrcCsQBOOd+AvwVuADYBjQD10Un0qOYchrExHnjhqacFu1oBszMWFacwbLiDP7jork8unofv311\nL1/943q+9ZdNXLQwj8uWFHDS1Ex8MRbtcEVknFIyFLa/oZXC9GC0wxARkRHinLv6OK874JMjFM7A\n+ZNh8sneuKFzvh7taIYkNSGOD72jhA+eXMza0noeeG0Pj64u46E3SslKiuf8+XlctDCPE0oyiFFi\nJCLDSMlQ2P6GVpYVp0c7DBERkf6bcQ48dQs0lEFKfrSjGTIzY1FRGouK0rjlonk8u6WSx9aW89Ab\ne7nvld3kJPu5YEEeFy/KY0lRuhIjERkyJUNAW2cXtc0dTFJZbRERGUumr/CSoW1Pw9Jrox3NsEqI\n93H+gjzOX5BHU1snf9tcyV/WlvHbV/fwy5d2kZ8aYMXcXE6bkc3J0zJJUkU6ERkE/c8BVGqOIRER\nGYty5kJyPmx9atwlQ70l+mO5ZFE+lyzKp7G1g6c37ecva8t56PVS7n15N3E+Y+nkdE6bmc1pM7KZ\nl5+iViMR6RclQxyaYygnxR/lSERERAagu8T2hj9CVwf44qIdUcQlB+K4bEkhly0ppK2zizd21fL8\n1gM8/1YV//3EFv77iS1kJsbzzhlZnDYjm9NmZpOdrPu7iPRNyRBeJTlQy5CIiIxB01fAm/dC6WtQ\nfEq0oxlR/lgfp0zP4pTpWdx8/mwqG1t5cesBXth6gBe2VvGn1WUALChI5YxZ2ZwxK5vFRemqTici\nPZQMcahlSGOGRERkzJl6BpjPGzc0wZKhI+UkB3jv0kLeu7SQUMixsbyBZ7dU8syWKu56Zht3/H0b\nqQlxnDYzmzNmqtVIRJQMAbC/sZV4XwxpwfHfvUBERMaZQCoUneSNGzr7lmhHM2rExBjzC1KZX5DK\np86aQV1zOy9sPcCzW6p47q0q/rzGazUqyQyydHI6S4rTWTo5jVm5yZqAXWQCUTIE7K9vJSfFj5ma\nzUVEZAyasQL+9nVo3A/JudGOZlRKC8Zz8aJ8Ll6U39Nq9OK2A7y52xtz9IdV+wAIxvtYVJjGsuJ0\nlhansaAgTa1HIuOYkiG8MUPqIiciImPW9HO8ZGj732DxNdGOZtTr3WoE4JyjtLaFN3bX8uYeb/nx\nc9vpCjkAcpL9zC9IZV5+SnhJpTA9QT+iiowDSobwusnNmZQS7TBEREQGZ9ICSMr1usopGRowM6Mo\nI0hRRpD3LCkAoLm9k3Wl9awva2DDvno2lDXw3FtVPQlSakIc8/JTWFCQypLJaSwuSmdSqn5YFRlr\nlAzhdZM7fWZ2tMMQEREZHDOvqtzmv0CoC2J80Y5ozAvGx3LS1ExOmprZ81xrRxebKxrZUFbP+n0N\nbCir5xf/2MX/PR8CvEJMi4vSWDw5jcVFaSwsTCUYr69aIqPZhP8XerCtk6b2LnWTExGRsW362bD6\nN7DvDSg6MdrRjEuBOJ+X7BSl9TzX1tnFxrIGVu+t61ke31ABgC/GmJ6dxPScJKblJDEtO5HpOUlM\nzUoiIV4Jq8hoMOGToe6y2ppjSERExrSpZ4LFeCW2lQyNGH+sjyWT01kyOb3nueqDbawprWP1njo2\nlHktSCvXlxPuYYcZFKQlMC2cKM2elMycvBSm5yQRiFOSJDKSlAzVe8lQTooqxYiIyBgWzICC5d64\noTO/Eu1oJrTMJD9nzc7lrNmHKvu1dnSxu7qZbZUH2VZ5kO1V3vqVHdW0dXrd7HwxxtSsRGbnpYQT\nJC9JmpQSULEGkQhRMtSoliERERknZpwDz/wnNB2AxKxoRyO9BOJ8zJqUzKxJyYc93xVy7KpuYnN5\nI5srGthU3siqPbU98yCBV+57ckaQksxEijODFPesg+SlJuCLUaIkMlhKhhraACVDIiIyDkxfAc98\nC7b/HRZeEe1opB98Mca07CSmZSdx4cK8nucbWjvYUtHI5vIGdh5oZnd1E1srG/n75krau0I9+8X7\nYijKSGBqdhJTsxKZmp3Ys52RGK8WJZHjmPDJUEV9K0n+WJL8E/6jEBGRsS5vMSTmwEt3wKwLwJ8U\n7YhkkFICcZxQksEJJRmHPd8VclQ0tLK7uond1c3sqm5i14EmdlQ18dyWqsMSpdSEOKaEE6TiDK81\naXJmkOKMoBIlkbAJnwFUNrZqvJCIiIwPMTFwyR3wwDXw4Ifg6t/9//buPDyu6r7/+Pur0TKjfZdt\nyRuy8ArYjtjMEsAOYS9pSCFtUkqgNE1TaJ6QlqS/BtI0LemShMY0CQkG0vCDpiEktAk/cAIUCARj\nbJnFSzE2xrK1S9a+6/z+OCNb3reRZkb383qe+9xFd66+cx9rjr9zzv0eSE2Pd1QSQ6EUozw/Qnl+\nhGWV+/9seMSxq62Xd5u72NbUzbamLrY3d/PKuy38dN2u/c7NyUhlemEms4ozmVGYxbT8MFNyw0zN\nizAlL0xRVjopGn4nARD4ZKiho19ltUVEZPKYexlcfS88+Vn4+Z/BR77nkySZ9EIpxoxo78/Fc/f/\nWd/gMLVtPbzX3MOO1ot23VEAACAASURBVB7eb+lmR2sPm+s6Wb2xgcFht9/5aSGjNCfM1LwwU/LC\nlOdHqCjMZHpBhOmFmZTnR1T5TiYFJUMdfQd1QYuIiCS1pZ+ErgZ49quQXQof/lq8I5I4C6eFmFOa\nw5zSnIN+NjLiaOkeoL69j7r2Xuo7+qhr76Oh3a/f2tXOM2837DcED6AsN4PpBZl7k6OyPN+7NCU3\nTFluBkXZGSruIAkv0MmQc47Gjn4NkxMRkcnngs9DVyO8shKyy+C82+IdkSSolBSjJCeDkpwMTqvI\nO+Q5IyOOxs5+drb1sLO1h52tvXu312xvpa69d+88SqNCKUZpTgZl0eSoJCeDkuzw3t9VnJ0eXWeo\nl0niJtDJUFvPIAPDIxomJyIik48ZXHYPdDfC6r/xPURn3BDvqCRJpaQYU6JD5g41omZoeGRv71J9\nRx+NHX5d395PY2cf25q6WbO9lbaewUNePzecyrT8CBUF/nmoioJMv13gtwsy01TwQcZFoJOhhg7N\nMSQiIpNYSop/ZqinxT8/lFkMVSviHZVMQqmhlGgPUJgzjnDewNAILd39NHX209zl102d/TR29rN7\nTy+1bb28uq2Vzv6h/V6XmR6iIDOdtJCRFkqJLmO2U1PITAsxJc8/5zQ1P8K0aPJWlhsmLaTn5uTQ\nAp0M1e9NhjRMTkREJqnUDLj+EXjoSl9h7sb/gorqeEclAZWemsLUvAhT8yJHPK+9d5Dath52tfkE\nqbatl/beQYZGRhgcHmFgyDE4POL3hxy9vYPU7enlN1ubD0qkzKA0J4MpuWHyM9PJjaSRF0klN5wW\n3U6LbqcyNS9MRUGmhu0FSKCToUb1DImISBCEc+EPfgIPfAge+Rjc/AwUV8U7KpHDyoukkRfJY+G0\nQz/DdCSdfYPUtfexe08v9e197G7vo26PLwyxp2eAHS3ddPQN0dE7yNCBDzpFleVmMLMwi+mFmcwo\nzGRmkS8UMb0gosIQk0ygk6GGjn4ASnLUMyQiIpNcThl88gl44FJ46Cr46A9g9gXxjkok5nLCaeSE\n0zi17ODKeWM55+gdHKajd4j23kHaewepa+9lR0sP77f28H5LD7/Z2szj0S/PR6UYFGVnUJabQWnO\naHGIMKU5GZTnR1gyI5/8TM3vlSwCnQzVd/RRmJVORqq6QkVEJACKKuHGJ+E/PgkPX+0rzl10J4TS\n4h2ZyIQzMzLTU8lMT2VK3uFHCY3O0bSjpYfde3pp7OynsaOfhs4+6tv7eKO2nZbuftyYTqa5ZTmc\nObuAs2YXcdaswiNeX+Ir0MlQY0cfpeoVEhGRIClbCH/yAjz1V/DiP8P2F3wvUcHMeEcmkpCONEfT\nqMHhEVq6BnivpZu177Xy6vZWnli3ix/99n0AZhRmcuasQs6eXcilC8vUc5RAAp0MNXT0K1MXEZHg\nyciGa++Dyovhvz8H3z0frv4WLPpovCMTSUppoZS9pcfPOaWIz+LLjW+s62DN9lbWbG/l2c0NPL6u\nli8/mcK1i8u5cdks5k/NjXfogRfwZKiPBfpHKCIiQXXadb6y3OO3wE8+Be8+C5f/I6RnxTsykaSX\nGkrh9Ip8Tq/I55YLTsE5x9u7O/jRb3fws5pdPPbaTs6aXcgfLZvFhxaUqfx3nAT2rg8Nj9Dc1a+y\n2iIiEmwFs+Cmp+CCO2D9I/C9C6FuQ7yjEpl0zIxF5Xnc89HT+e0Xl/OlK+axe08vn3lkHRd8/TlW\nPvsOzV398Q4zcAKbDDV3DTDioEzD5EREJOhCabD8b3xxhYFu+P5y+MXnoWN3vCMTmZTyM9O59cJK\n/ucLF/ODP6ymqiybf37mf1n2D8/yxz9cyw9feY/tzd04d+jS3xI7gR0m1zA6x1COkiEREREAZl8I\nn/4NPPtVeP0hWPfvcOYtcP7nILsk3tGJTDqhFGPFgjJWLChja2MXP/rtDn61qYHVGxsAKM+PcEFV\nMedXFXNeZTEFWSq8EGvjmgyZ2WXAvUAI+IFz7p7DnPdR4CfAmc65teMZ06h6TbgqIiJysKwiX0zh\nvNvhhX+CV78Drz8IZ/8JLLsNMgvjHaHIpDSnNJu7r1nIXVcvYEdLDy9ubeald5r4xZt1PPbaTsxg\n0bQ8zjmlkKqyHOaUZjOnNJvcsErjn4xxS4bMLATcB3wIqAVeM7MnnXMbDzgvB7gdeHW8YjmUxtFk\nKE/PDImIiBykcDZc+2++V+j5e+Clb8FrD8A5n4FzPwPhvHhHKDIpmRmzirOYVZzFJ8+ZydDwCG/s\naueld5p56Z1mHn55BwPDI3vPL8vN8IlRiU+OKkuzqSzJpjQnAzOL4ztJDuPZM3QWsNU5tw3AzB4D\nfgfYeMB5XwW+DnxhHGM5SENHP6EUoyhLyZCIiMhhFVfBdQ/4CVqf/3v4n3vg1e/Cwo9A5SV+aF0k\nP95RikxaqaEUls4oYOmMAm5bXsXQ8Ag723rZ2ti1b2nq4vF1u+jqH9r7uuyMVGYXZzG7OItTSvy6\nsiSb2cVZZGUE9kmZg4znnSgHdo7ZrwXOHnuCmS0FpjvnfmFmh02GzOxW4FaAGTNmxCS4+o4+SrIz\nCKUoYxYRETmqsgVw/Y9gdw289A148yd++JylQHm1T4wqL4HyD0BI/9ESGS+poZS9Sc6HFpTtPe6c\no76jj62NXWxv7mZbUzfbmrtZ934b//XGbsbWYsjOSCU/M42CzHQKstIpiG6PHivLDVNZksWMokwy\nUkNxeJcTJ26fVmaWAnwD+KOjneucux+4H6C6ujomZTUaOvpUSU5Ejtvg4CC1tbX09fXFO5QJEQ6H\nqaioIC1NY9Ilatpi+L0fwvAg1K71cxO9+yy88I++1ygj1/cWzb/a9x6lagSGyEQwM6bmRZiaF+GC\nqv0LnvQNDrOjpYdtTV1sa+6muaufPT2DtPUM0NY9wHvN3bT1DNDZN7Tf61IMKgoy9/YsnVKSTWV0\nXZY7OYbhjWcytAuYPma/InpsVA6wCHg+eiOnAE+a2TUTUUShsaOfmUWZ4/1rRGSSqa2tJScnh1mz\nZk2KRuBInHO0tLRQW1vL7Nmz4x3OuDhaoR8zmwE8DORHz7nTOffLCQ80EYXSYOa5frnkr6GnFba/\nsC852vzf8MzfwJk3Q/WnILs03hGLBFY4LcTcKTnMnZJzxPMGh0fY0zNIXXuv71mKJk/bmrp5dVsr\nvYPDe8/Ni6Qxtyxn73XnTsnh1LIc8iLJ9eXZeCZDrwFVZjYbnwTdAPz+6A+dc+1A8ei+mT0P3DFR\n1eQaOvs4a7Yq4ojI8enr6wtEIgT+W8aioiKampriHcq4OMZCP/8H+LFz7jtmtgD4JTBrwoNNBpmF\nsPBavzjnE6JXvwvP/wO8+C+w6KNw9qd9z5KIJKS0UAolORmU5GRwesX+zwKODsPb3tTN1qYuNtd3\nsqW+k5+t30XnmGeVpuWFqSrLoSw3g4KsdIqy0inITKcwyw/JK8xMpzA7nZyM1IRoS8ctGXLODZnZ\nZ4Gn8d+mrXLOvW1mfwusdc49OV6/+2j6BofZ0zNIWa667kXk+CXCh/dEmeTv9VgK/TggN7qdB2gW\n0mNhBnOW+6V5K6z5Hqx/BDY8CjPO9UnRvKv0bJFIEhk7DG/ZnL39GTjn2N3ex5b6DjbXd/K/9Z28\n09jFlvpOWrsH9qt8N1ZmeojpBZlUFESYXrhvPb0gk+mFEXImqGT4uH4KRYcS/PKAY18+zLkXjWcs\nYzV29AOaY0hEJOCOWugHuBt4xsz+HMgCVhzqQuNR6GfSKJ4DV/wTXPzXsP5HPjH6zxshbwac86ew\n9A8hIzveUYrICTIzyvMjlOdHuGRe2X4/c87RPTBMW/cArQcsu9t72dnaS21bD69ub92vEh7ArKJM\nnv/CxeMefyC/kmno1ISrIpKcWlpaWL58OQD19fWEQiFKSvyDsmvWrCE9/eizk990003ceeedzJ07\nd1xjnSQ+DjzknPsXMzsX+HczW+Sc2++rzvEo9DPpRPJh2Wd9ArTlKXjlPnj6i77oQvXNvrcop+zo\n1xGRpGFmZGekkp2RyvTCwz+r75xjT88gO9t62Nnay862HoZHJuajNJDJUH27kiERSU5FRUXU1NQA\ncPfdd5Odnc0dd9yx3znOOZxzpKSkHPIaDz744LjHmSSOVugH4GbgMgDn3CtmFsY/79o4IRFORikh\nmH+VX3a+Bi/fCy99E15ZCWfcAOf+OZScGu8oRWQCmZkv8Z2VftCzSuMtkMlQQ4dPhqYoGRKRk/CV\n/3qbjbs7YnrNBdNyuevqhcf9uq1bt3LNNdewZMkS1q9fz+rVq/nKV77CunXr6O3t5frrr+fLX/aj\nlM8//3xWrlzJokWLKC4u5tOf/jRPPfUUmZmZ/PznP6e0NDBVv45Y6CfqfWA58JCZzQfCwOSsKBEP\n08/0cxe1vOuTofWPwLofwtwrYNltMOMc//yRiMg4OfTXhpNcY2c/Gakp5EYCmQuKyCS1efNmPve5\nz7Fx40bKy8u55557WLt2LRs2bGD16tVs3LjxoNe0t7fzwQ9+kA0bNnDuueeyatWqOEQeH865IWC0\n0M8mfNW4t83sb83smuhpnwf+2Mw2AI8Cf+Sc0zC4WCuqhKu+CZ97Gy78S3j/FXjwMvjOMj+crrs5\n3hGKyCQVyGygvr2PstzwZK+SJCLj7ER6cMZTZWUl1dXVe/cfffRRHnjgAYaGhti9ezcbN25kwYIF\n+70mEolw+eWXA/CBD3yAF198cUJjjrejFfqJltk+b6LjCqzsEj9n0fl/AW/8hy+48PSXYPWX4dTL\nYMknYc4KVaETkZgJ5KdJQ0efhsiJyKSTlZW1d/udd97h3nvvZc2aNeTn5/OJT3yCvr6+g14ztuBC\nKBRiaGjooHNEJlx6lp+otfpT0LjJJ0UbHvMTuWaXwRkfhyWfgOKqeEcqIkkusMPkSjXHkIhMYh0d\nHeTk5JCbm0tdXR1PP/10vEMSOTGl8+HDX4PPb4brH4HyD8DL34aV1fBvy+Dpv4atv4KBnnhHKiJJ\nKHA9Q8456tv7uGReYB4QFpEAWrp0KQsWLGDevHnMnDmT887TSC9JcqG0fVXoOhvgzR/DO8/Amvt9\n8YVQhi+4UHkxVF4CZafBYSoqioiMsmR7DrS6utqtXbv2hF/f0TfI6Xc/w5eumMetF1bGMDIRCYJN\nmzYxf/78eIcxoQ71ns3sdedc9WFeEmgn207JcRrogfdfhnef80vj2/54ZjHMXAblS2HaEpi62M91\nJCKBcKztVOB6hho7NMeQiIjIpJGe6YsqzFnh9zvrYdvz8O6z8P5vYdOT+84tPMUnRqNLyTwYGYLB\nHhjsjS5jtocH/GvKFkGa/t8gMhkFLhlq6OgHlAyJiIhMSjlT/OStZ9zg93taYfd6qKvx651r4K3H\nj++aKWlQtgCmLd3X01Qy/+CqdiMj0NsG3Y3Q3QRdjX6epOnnQF55bN6fiMRU4JKh+nb1DImIiARG\nZiHMWe6XUV1NPjlq2QqhdEjLhLTIweuUVGjaDLvXwa518NZP4fUH/TVSIzD1dEjP9slPVxP0NPue\npkMpmAUzz4suy/y+pvgQibvAJUMNnaPJkKrJiYiIBFJ2CVR9yC9HU3IqLIjOwTsyAq3bfHK0e71P\nkHrbIGcaTD0DskohuxSySvySXQpDfbDjFdjxG9jyFNQ84q+VW+6Touln+4QtLcsP+du7zvQlxtMi\n0NcBXQ1+6az3PU5d9fu207P8tWYu89X20iLjd+9EJpnAJUONHf3khFPJTA/cWxcREZGTkZICxXP8\ncvrvHfvrpi2Bcz/jk6nmLfDeS7DjZdj+Arz5nycWSzjfDwnMLoXuZnju7wHne7rKPwAzzvW9UNPP\ngnDuvtc5BwPd0N/pl4FO/3xU8an+WiIBE7iMoL69T0PkREREZOKlpPh5k0rnw1l/7BOTjt3Q3+Gr\n4g12H7COLhm5frLZ0eQnuwxSDxjh0tsG77/qe6B2vAwv/yu89A2wFD8kb6h/XwLEYSoJ58+A8mqo\nOBMqqmHK6eNfOGJ4CNrfh7YdMOU0yCoe398ncoDAJUMNnX1MUTIkIkmqpaWF5cv9sw/19fWEQiFK\nSkoAWLNmDenp6cd0nVWrVnHFFVcwZcqUcYtVRI7CLFpYIQbFFSIFMPcyv4Dv/al9zSdGTVv8s00Z\nOZAxus7xSVZ6tp/DqXGjP3/nGnj7p/4aKWk+QamohswiX11vqH/MehCG+/12SggihX7IX6RgzHZ0\nnRqGPTv8c1otW6HlXb9u3Q4jg9HflwpVl/riF6dednDCd6Kc88Mbm7b4IYhjY0vPjM3vkKQVuGSo\nsaOfs0/JincYIiInpKioiJqaGgDuvvtusrOzueOOO477OqtWrWLp0qVKhkQmq/QsOOUivxyLsQUm\nOupg11qoXQu7Xof1j/jeKgv5BCWU5ie5Hbvthn3lvt42v30koQwoqoSSuTDvSiiaA7nTfEn0N34M\nW37phwEu+l044/d9MnasxSacgz3v+2e6Rpe6GuhrP/T5qeH9k7iiOTBtsX8GrHQhpB7bF0ySvAKV\nDI2MOBo6NExORGLkqTuh/s3YXnPKaXD5PSf00ocffpj77ruPgYEBli1bxsqVKxkZGeGmm26ipqYG\n5xy33norZWVl1NTUcP311xOJRI6rR0lEAiB3KuReDfOv9vsjI4DzvT9HMzLih/31tkJPW3Td6of7\nFcyMJj4VfsjggSovgeV3+aRow2NQ8yisXQWFlXDGx32CMtDle70Gug/e7qyD3TX+d0K0JPpCWPi7\n/rmt0gW+J6undV9cY+Psbt6/YuBoSfWpi/dPkE5m6OBAjy9woUqCCSNQyVBrzwBDI07D5ERk0nnr\nrbd44oknePnll0lNTeXWW2/lscceo7KykubmZt580ydte/bsIT8/n29/+9usXLmSxYsXxzlyEUl4\nh0pcjnRuJN8vhSfyu0L7SqH3dfhJczc8Bs/93WHOT/VD/dKzfe/OvCv3TapbtvD4h9o5B23bfVJV\nV+PXG38O6x72P7cUn5yVzvNzTY0+A1Y0x/eSjervhMbN0LQJGjf5YYiNm30VwMJKH+e8q/zzWcdz\nfyXmApUMNXSorLaIxNAJ9uCMh1/96le89tprVFdXA9Db28v06dP58Ic/zJYtW7jtttu48sorufTS\nS+McqYjIMQrnwpJP+KW91g/fy8j2QwDTo+tQemx7Wcyg8BS/LPpdf8w5/7zT7hpoeCua3GyCzb8A\nN+LPSUmFoipf5KLlXV8UYlRqxCdPc5ZD/kzY+Sr89ju+yEVWKcy93CdHsz94cK9Tdws0/6+vQtj8\njt8e6oNZF/ghkNOWHjz5rxyXQN29xo5+AErVMyQik4xzjk996lN89atfPehnb7zxBk899RT33Xcf\njz/+OPfff38cIhQROQl5FX6JBzNfka9gFiy8dt/xwT6fnDRt3r/nZ8bZUHKjH5ZXOt8nQAf2/vS1\nwzurfUL11k99z1NaFlStgHCeT3yatuwb8gf++aaiKjB8KfXnvuaLYIwmRqdcBMVVsUkOB/v8JMLd\nzdF1y5j9Fj8sMTUSnRMrcvD8WOlZ/lmsrBJfITCcn7A9YIFKhur39gwpGRKRyWXFihVcd9113H77\n7RQXF9PS0kJ3dzeRSIRwOMzHPvYxqqqquOWWWwDIycmhs7MzzlGLiCSxtDBMPd0vxyucB6dd55eh\nftj+Imz5hZ+Yd3gAiuf6yX6LT40uVZA3Y19C0d0C773gn6969zn/WvCT+c4635dJH50EOLts3zoj\n25/nHHQ3+Wp+rdv80jZmu7ft0HGnpPrKgunZvodqoNs/DzY8cOT3ayH/uqwSyIqu0yJ+2KGF/Dol\nNGY7xSdQFx5/gaDjFahkaHSYXGmOhsmJyORy2mmncdddd7FixQpGRkZIS0vju9/9LqFQiJtvvhnn\nHGbG17/+dQBuuukmbrnlFhVQEBGJt9QM3yNUtQKu+uaxvSarCBZ+xC/gk5ptz/tl+wvQ1bBvCN9Y\naZm+p6an1ffujLIU3/NWeIq/Zm65Py+zeMy6yCcoh+p5Gh7af36sgS7fg9Td4pOu0V6l0Z6m3et9\nEjgy7KsPupHottu3nzNlQpIhc+4wE28lqOrqard27doTeu3P1u/i+S2NfOuGJTGOSkSCYtOmTcyf\nPz/eYUyoQ71nM3vdOVcdp5AS2sm0UyIiMTEy7JORrkafGI1ddzf5YhMFs/c9H5U/Y9KVET/WdipQ\nPUPXLinn2iUxmNhMRERERCRRpYSiQ+NKgUXxjiahJeaTTCIiIiIiIuNMyZCIyHFKtuHFJyNI71VE\nRIJHyZCIyHEIh8O0tLQEIklwztHS0kI4rAqcIiIyOQXqmSERkZNVUVFBbW0tTU1N8Q5lQoTDYSoq\n4jS3h4iIyDhTMiQichzS0tKYPXt2vMMQERGRGNAwORERERERCSQlQyIiIiIiEkhKhkREREREJJAs\n2SoimVkTsOMkLlEMNMconPGUDHEqxthJhjiTIUZIjjgnQ4wznXMlExVMMlE7lVAUY+wkQ5zJECMk\nR5yTIcZjaqeSLhk6WWa21jlXHe84jiYZ4lSMsZMMcSZDjJAccSpGOZJkuffJEKdijJ1kiDMZYoTk\niDNIMWqYnIiIiIiIBJKSIRERERERCaQgJkP3xzuAY5QMcSrG2EmGOJMhRkiOOBWjHEmy3PtkiFMx\nxk4yxJkMMUJyxBmYGAP3zJCIiIiIiAgEs2dIREREREREyZCIiIiIiARToJIhM7vMzLaY2VYzuzPe\n8RyKmb1nZm+aWY2ZrY13PKPMbJWZNZrZW2OOFZrZajN7J7ouSMAY7zazXdH7WWNmV8Q5xulm9pyZ\nbTSzt83s9ujxRLuXh4szYe6nmYXNbI2ZbYjG+JXo8dlm9mr07/w/zCw9AWN8yMy2j7mPi+MV45hY\nQ2a23sz+O7qfMPcxSNROnTi1UzGLUe1U7GJUOxVj49FWBSYZMrMQcB9wObAA+LiZLYhvVId1sXNu\ncYLVd38IuOyAY3cCv3bOVQG/ju7H00McHCPAN6P3c7Fz7pcTHNOBhoDPO+cWAOcAfxb9d5ho9/Jw\ncULi3M9+4BLn3BnAYuAyMzsH+Ho0xjlAG3BzAsYI8IUx97EmfiHudTuwacx+It3HQFA7ddIeQu1U\nLKidih21U7EX87YqMMkQcBaw1Tm3zTk3ADwG/E6cY0oazrkXgNYDDv8O8HB0+2Hg2gkN6gCHiTGh\nOOfqnHProtud+D/ochLvXh4uzoThvK7oblp0ccAlwE+ix+N6L48QY0IxswrgSuAH0X0jge5jgKid\nOglqp2JD7VTsqJ2KrfFqq4KUDJUDO8fs15JgfzRRDnjGzF43s1vjHcxRlDnn6qLb9UBZPIM5gs+a\n2RvR4Qlx7dYfy8xmAUuAV0nge3lAnJBA9zPaXV4DNAKrgXeBPc65oegpcf87PzBG59zoffxa9D5+\n08wy4hgiwLeAvwRGovtFJNh9DAi1U7GXsJ+tB0iYz9Wx1E6dPLVTMTUubVWQkqFkcb5zbil+mMSf\nmdmF8Q7oWDhfoz0Rv0n4DlCJ7/qtA/4lvuF4ZpYNPA78hXOuY+zPEuleHiLOhLqfzrlh59xioAL/\nrfq8eMZzKAfGaGaLgC/iYz0TKAT+Kl7xmdlVQKNz7vV4xSBJR+1UbCXU5+ootVOxoXYqNsazrQpS\nMrQLmD5mvyJ6LKE453ZF143AE/g/nETVYGZTAaLrxjjHcxDnXEP0j3wE+D4JcD/NLA3/wf2Ic+6n\n0cMJdy8PFWci3k8A59we4DngXCDfzFKjP0qYv/MxMV4WHd7hnHP9wIPE9z6eB1xjZu/hh2VdAtxL\ngt7HSU7tVOwl3GfrgRLxc1XtVOypnTpp49ZWBSkZeg2oiladSAduAJ6Mc0z7MbMsM8sZ3QYuBd46\n8qvi6kngxuj2jcDP4xjLIY1+cEd9hDjfz+j41geATc65b4z5UULdy8PFmUj308xKzCw/uh0BPoQf\nM/4ccF30tLjey8PEuHnMfygMP745bvfROfdF51yFc24W/nPxWefcH5BA9zFA1E7FXkJ9th5KIn2u\ngtqpWFI7FTvj2VaZ7+kMBvPlFb8FhIBVzrmvxTmk/ZjZKfhv2QBSgf+bKDGa2aPARUAx0ADcBfwM\n+DEwA9gB/J5zLm4Phh4mxovwXeUOeA/4kzFjniecmZ0PvAi8yb4xr1/Cj3NOpHt5uDg/ToLcTzM7\nHf+wZAj/xc6PnXN/G/07egzfrb8e+ET0m61EivFZoAQwoAb49JgHWOPGzC4C7nDOXZVI9zFI1E6d\nOLVTsaF2KqYxqp0aB7FuqwKVDImIiIiIiIwK0jA5ERERERGRvZQMiYiIiIhIICkZEhERERGRQFIy\nJCIiIiIigaRkSEREREREAknJkMgJMrNhM6sZs9wZw2vPMrNEnrtDRESSgNoqkSNLPfopInIYvc65\nxfEOQkRE5AjUVokcgXqGRGLMzN4zs380szfNbI2ZzYken2Vmz5rZG2b2azObET1eZmZPmNmG6LIs\neqmQmX3fzN42s2eiM0OLiIicNLVVIp6SIZETFzlg6MH1Y37W7pw7DViJn00e4NvAw86504FHgH+N\nHv9X4H+cc2cAS4G3o8ergPuccwuBPcBHx/n9iIjI5KO2SuQIzDkX7xhEkpKZdTnnsg9x/D3gEufc\nNjNLA+qdc0Vm1gxMdc4NRo/XOeeKzawJqHDO9Y+5xixgtXOuKrr/V0Cac+7vxv+diYjIZKG2SuTI\n1DMkMj7cYbaPR/+Y7WH0jJ+IiMSW2ioJPCVDIuPj+jHrV6LbLwM3RLf/AHgxuv1r4E8BzCxkZnkT\nFaSIiASa2ioJPGXvIicuYmY1Y/b/n3NutGRpgZm9gf/G7OPRY38OPGhmXwCagJuix28H7jezm/Hf\nqv0pUDfu0YuISBCorRI5Aj0zJBJj0XHY1c655njHIiIicihqq0Q8DZMTEREREZFAUs+QiIiIiIgE\nknqGREREREQkOVVfggAAACxJREFUkJQMiYiIiIhIICkZEhERERGRQFIyJCIiIiIigaRkSERERERE\nAun/A6YVFRWmy8vXAAAAAElFTkSuQmCC\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "The confusion matrix:\n", "\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0cAAAEYCAYAAACX/Y2IAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMzMD3DgyMzMzMzMD3DgyMzMzMzMD4P8BzzR5rYUM8bYAAAAASUVORK5CYII=\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "\n", "The classification report:\n", "\n", " precision recall f1-score support\n", "\n", " Neutral 0.63 0.66 0.64 3500\n", " Happy 0.90 0.83 0.86 3500\n", " Sad 0.67 0.75 0.71 3500\n", " Surprised 0.82 0.83 0.83 3205\n", " Angry 0.80 0.72 0.76 3500\n", "\n", " accuracy 0.76 17205\n", " macro avg 0.76 0.76 0.76 17205\n", "weighted avg 0.76 0.76 0.76 17205\n", "\n" ] } ], "source": [ "# Training the model (5 Classes):\n", "start_time = time.time()\n", "history = model.fit(X_train, y_train, validation_data = (X_val, y_val), \n", " batch_size = mini_batch_size, \n", " epochs = num_epochs, verbose = 2)\n", "end_time = time.time()\n", "\n", "# Creating a new directory to save model and history files:\n", "path = root_dir + '/Models/' + str(model_number)\n", "if not os.path.exists(path):\n", " os.mkdir(path)\n", "\n", "model.save(path + os.sep + 'model.h5') # Saving the model\n", "json.dump(history.history, open(path + os.sep + 'Model_history.json', 'w')) # Saving the history\n", "plot_model(model, to_file=path + os.sep + 'Model.png', show_shapes = True) # Saving the model visualization\n", "\n", "# Updating the Excel sheet:\n", "row_index = params_sheet[params_sheet['model'] == model_number].index[0]\n", "params_sheet.loc[row_index, 'train_acc'] = round(max(history.history['acc'])*100, 2)\n", "params_sheet.loc[row_index, 'val_acc'] = round(max(history.history['val_acc'])*100, 2)\n", "params_sheet.loc[row_index, 'train_time'] = round((end_time - start_time)/60)\n", "params_sheet.to_excel(root_dir + '/Model Params.xlsx')\n", "\n", "# Plotting training history:\n", "summary_visualizer(history, path, X_test, y_test, params_sheet, model_number)" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "5ABu9fiMaed1" }, "source": [ "One interesting observation above is that for both 3-class and 5-class models, the validation accuracy (test on the plot) is always higher than the training accuracy, although a reversed trend is expected in most model training processes (similar observation with the loss values). This is an expected trend from the model training process on Keras for the following two reasons (source: [Keras FAQ](https://keras.io/getting-started/faq/#why-is-the-training-loss-much-higher-than-the-testing-loss)):\n", "\n", "1. Regularization mechanisms, such as Dropout and L1/L2 weight regularization, are turned off at testing/validation time.\n", "\n", "2. The training loss is the average of the losses over each batch of training data. Because the model is changing over time, the loss over the first batches of an epoch is generally higher than over the last batches. On the other hand, the testing/validation loss for an epoch is computed using the model as it is at the end of the epoch, resulting in a lower loss.\n", "\n", "### 1.4. Performance Analysis:\n", "\n", "As the last section of this notebook, we look back at the human accuracies that we defined in the fist notebook (EDA) and compare our model accuracies with the accuracies that we got from hand-labelling a set of 1000 sample images for 5 classes and 600 sample images for 3 classes:" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 295 }, "colab_type": "code", "executionInfo": { "elapsed": 396, "status": "ok", "timestamp": 1560306979273, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "vlcUVJB0bq22", "outputId": "03174999-a0d3-4c08-ca6c-17c66195482b" }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUg4WZmZWysHCzMxK/QCsHhiFdSZplAAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" } ], "source": [ "# Loading the data for 3 Classes:\n", "model_number = 9\n", "with open(root_dir + '/Models/9/Model_history.json') as json_file: \n", " history = json.load(json_file)\n", " \n", "plt.plot(history['val_acc'])\n", "plt.title('Model Accuracy (3-Class)')\n", "plt.ylabel('Validation Accuracy')\n", "plt.xlabel('Epoch')\n", "plt.axhline(0.8483, color = 'red', ls = ':', linewidth=2)\n", "plt.text(11.5, 0.83, 'Human Accuracy = 84.8%', color = 'red', size = 14)\n", "plt.text(12.3, 0.915, 'Model Accuracy = 90.2%', color = '#186aa3', size = 14)\n", "plt.ylim((0.7, 0.95));" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 295 }, "colab_type": "code", "executionInfo": { "elapsed": 436, "status": "ok", "timestamp": 1560306857216, "user": { "displayName": "Arash Tavassoli", "photoUrl": "https://lh5.googleusercontent.com/-TL3G5GlWXr4/AAAAAAAAAAI/AAAAAAAAEUk/2aiZ3iwrMDo/s64/photo.jpg", "userId": "00927822409229306590" }, "user_tz": 420 }, "id": "DpBdeFxlZEHX", "outputId": "2e1fc816-5eb8-4280-949f-c887de511e64", "scrolled": true }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQSYLSdIgk4UkaZDJQpI0yGQhSRr0A+i94MUmFGnZAAAAAElFTkSuQmCC\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" } ], "source": [ "# Loading the data for 5 Classes:\n", "model_number = 11\n", "with open(root_dir + '/Models/11/Model_history.json') as json_file: \n", " history = json.load(json_file)\n", " \n", "plt.plot(history['val_acc'])\n", "plt.title('Model Accuracy (5-Class)')\n", "plt.ylabel('Validation Accuracy')\n", "plt.xlabel('Epoch')\n", "plt.axhline(0.6930, color = 'red', ls = ':', linewidth=2)\n", "plt.text(16.68, 0.775, 'Model Accuracy = 75.9%', color = '#186aa3', size = 14)\n", "plt.text(15.5, 0.67, 'Human Accuracy = 69.3%', color = 'red', size = 14)\n", "plt.ylim((0.54, 0.82));" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "Rr_8yAvkwVVl" }, "source": [ "\n", "\n" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "Ncjwg89vwVVl" }, "source": [ "Although the validation accurcies of 90.2% and 75.9% for the 3-class and 5-class models may not sound super impressive, it is shown that the models are still performing superior to what a human can do on the same set of data. This is once again explained by the fact that many images in this dataset are difficult to classify, even for a human mind.\n", "\n", "We can now use the saved models in the next notebook to build a real-time facial expression recognizer!" ] } ], "metadata": { "accelerator": "GPU", "colab": { "collapsed_sections": [], "name": "Part 3 - Model Training and Analysis.ipynb", "provenance": [], "version": "0.3.2" }, "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.2" } }, "nbformat": 4, "nbformat_minor": 1 }