GradientAttack.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "id": "initial_id",
   "metadata": {
    "collapsed": true,
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.246769Z",
     "start_time": "2024-08-28T08:44:57.885177Z"
    }
   },
   "source": [
    "import torch\n",
    "from torch.optim import Optimizer\n",
    "from torchvision import models, datasets, transforms\n",
    "from torch.utils.data import DataLoader\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import utils_a\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import torch.optim.lr_scheduler as lr_scheduler\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "utils_a.set_seed(42)\n",
    "device = utils_a.set_device()"
   ],
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using mps device.\n"
     ]
    }
   ],
   "execution_count": 1
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.300747Z",
     "start_time": "2024-08-28T08:45:01.248276Z"
    }
   },
   "cell_type": "code",
   "source": [
    "dataset_folder = \"mri_brain_tumor\"\n",
    "#mean, std = utils_a.compute_mean_std(train_loader)\n",
    "mean, std = torch.tensor([0.1855] * 3), torch.tensor([0.1813] * 3)  # precomputed, to save time\n",
    "\n",
    "print(f\"Mean: {mean}\")\n",
    "print(f\"Std: {std}\")\n",
    "train_transform = transforms.Compose([\n",
    "    transforms.RandomHorizontalFlip(p=0.5),  # Flip orizzontale\n",
    "    transforms.RandomVerticalFlip(p=0.5),  # Flip verticale\n",
    "    transforms.RandomRotation(degrees=15),  # Rotazione casuale di 15 gradi\n",
    "    transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2, hue=0.1),  # Variazione casuale dei colori\n",
    "    transforms.RandomResizedCrop(size=224, scale=(0.8, 1.0)),  # Crop e ridimensionamento casuale\n",
    "    transforms.ToTensor(),  # Converti in tensore\n",
    "    transforms.RandomErasing(p=0.5, scale=(0.02, 0.2), ratio=(0.3, 3.3)),  # Erasing casuale\n",
    "    transforms.Normalize(mean=mean, std=std)  # Normalizzazione\n",
    "])\n",
    "\n",
    "# Transformation for testing dataset without data augmentation\n",
    "test_transform = transforms.Compose([\n",
    "    transforms.Resize((224, 224)),  # Resize the image to 224x224\n",
    "    transforms.ToTensor(),  # Convert the image to a tensor\n",
    "    transforms.Normalize(mean=mean, std=std)  # Normalize the image\n",
    "])\n",
    "\n",
    "# Apply the transformations to the datasets\n",
    "train_dataset = datasets.ImageFolder(root=f'{dataset_folder}/Training', transform=train_transform)\n",
    "test_dataset = datasets.ImageFolder(root=f'{dataset_folder}/Testing', transform=test_transform)\n",
    "\n",
    "train_loader = DataLoader(train_dataset, batch_size=8, shuffle=True, num_workers=1)\n",
    "test_loader = DataLoader(test_dataset, batch_size=8, shuffle=True, num_workers=1)"
   ],
   "id": "511724dc7cafe137",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean: tensor([0.1855, 0.1855, 0.1855])\n",
      "Std: tensor([0.1813, 0.1813, 0.1813])\n"
     ]
    }
   ],
   "execution_count": 2
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.303873Z",
     "start_time": "2024-08-28T08:45:01.301765Z"
    }
   },
   "cell_type": "code",
   "source": "num_epochs = 5",
   "id": "5030255c897110db",
   "outputs": [],
   "execution_count": 3
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "#### ATTACK 5: manipulating the gradient direction",
   "id": "d9042f0f94efcb7"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.312529Z",
     "start_time": "2024-08-28T08:45:01.305881Z"
    }
   },
   "cell_type": "code",
   "source": [
    "class DrunkOptimizer(Optimizer):\n",
    "    def __init__(self, params, base_optimizer, scale_strength=0.7, perturb_strength=0.7, attack_type='combined', target_layers=None):\n",
    "        defaults = dict(lr=0.001, weight_decay=0.9)\n",
    "        super(DrunkOptimizer, self).__init__(params, defaults)\n",
    "        self.base_optimizer = base_optimizer\n",
    "        self.scale_strength = scale_strength\n",
    "        self.perturb_strength = perturb_strength\n",
    "        self.attack_type = attack_type\n",
    "        self.target_layers = target_layers\n",
    "        self.optimizer = base_optimizer\n",
    "\n",
    "    def step(self, closure=None):\n",
    "        for group in self.optimizer.param_groups:\n",
    "            for param in group['params']:\n",
    "                if param.grad is not None:\n",
    "                    self.attack_gradients(param)\n",
    "        self.optimizer.step(closure)\n",
    "    \n",
    "    def attack_gradients(self, param):\n",
    "        if self.target_layers is None or any(name in param.name for name in self.target_layers):\n",
    "            if self.attack_type == 'scale':\n",
    "                param.grad.data *= (1 - self.scale_strength)\n",
    "            elif self.attack_type == 'perturb':\n",
    "                noise = torch.randn_like(param.grad.data) * self.perturb_strength\n",
    "                param.grad.data += noise\n",
    "            elif self.attack_type == 'combined':\n",
    "                param.grad.data *= (1 - self.scale_strength)\n",
    "                noise = torch.randn_like(param.grad.data) * self.perturb_strength\n",
    "                param.grad.data += noise\n",
    "                \n",
    "    def zero_grad(self):\n",
    "        self.base_optimizer.zero_grad()"
   ],
   "id": "916e42bb7ee25b4e",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.315304Z",
     "start_time": "2024-08-28T08:45:01.313569Z"
    }
   },
   "cell_type": "code",
   "source": "",
   "id": "1fb6c8f0571f5739",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### DIFESA",
   "id": "7559d55697236875"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.318815Z",
     "start_time": "2024-08-28T08:45:01.316558Z"
    }
   },
   "cell_type": "code",
   "source": "",
   "id": "6e49e4468608dd3",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T08:45:01.326442Z",
     "start_time": "2024-08-28T08:45:01.319946Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Function to train and evaluate a model, returning accuracy data\n",
    "def train_and_evaluate(model, optimizer, train_loader, test_loader, name, num_epochs=5, adv_training=False):\n",
    "    criterion = nn.CrossEntropyLoss()\n",
    "    scheduler = lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', factor=0.1, patience=5, verbose=True)\n",
    "\n",
    "    train_accuracies = []\n",
    "\n",
    "    for epoch in range(num_epochs):\n",
    "        model.train()\n",
    "        correct = 0\n",
    "        total = 0\n",
    "\n",
    "        for inputs, labels in train_loader:\n",
    "            inputs, labels = inputs.to(device), labels.to(device)\n",
    "            optimizer.zero_grad()\n",
    "            outputs = model(inputs)\n",
    "            loss = criterion(outputs, labels)\n",
    "            loss.backward()\n",
    "\n",
    "            if adv_training:\n",
    "                for param in model.parameters():\n",
    "                    if param.grad is not None:\n",
    "                        param.grad.data += 0.01 * torch.randn_like(param.grad.data)\n",
    "\n",
    "            optimizer.step()\n",
    "\n",
    "            _, predicted = outputs.max(1)\n",
    "            total += labels.size(0)\n",
    "            correct += predicted.eq(labels).sum().item()\n",
    "\n",
    "        epoch_accuracy = 100.0 * correct / total\n",
    "        train_accuracies.append(epoch_accuracy)\n",
    "\n",
    "        print(f'Epoch {epoch+1}/{num_epochs}, Accuracy: {epoch_accuracy:.2f}%')\n",
    "        scheduler.step(epoch_accuracy)\n",
    "\n",
    "    all_labels, all_preds = utils_a.test(model, device, test_loader, name)\n",
    "    print(classification_report(all_labels, all_preds, target_names=train_dataset.classes))\n",
    "    \n",
    "    return train_accuracies"
   ],
   "id": "79fa601fae413c24",
   "outputs": [],
   "execution_count": 5
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T09:26:29.382106Z",
     "start_time": "2024-08-28T08:45:01.327500Z"
    }
   },
   "cell_type": "code",
   "source": [
    "print(\"Training the clean model...\")\n",
    "clean_model = utils_a.CNN(num_classes=4).to(device)\n",
    "state_dict = torch.load('weights/cnn_10epochs.pth', map_location=device, weights_only=True)\n",
    "clean_model.load_state_dict(state_dict)\n",
    "clean_optimizer = optim.Adam(clean_model.parameters(), lr=0.0001, weight_decay=1e-5)\n",
    "clean_accuracies = train_and_evaluate(clean_model, clean_optimizer, train_loader, test_loader, \"clean\", num_epochs)\n",
    "\n",
    "print(\"Training the model attacked with DrunkOptimizer...\")\n",
    "attacked_model = utils_a.CNN(num_classes=4).to(device)\n",
    "attacked_model.load_state_dict(state_dict)\n",
    "base_optimizer = optim.Adam(attacked_model.parameters(), lr=0.0001, weight_decay=1e-5)\n",
    "drunk_optimizer = DrunkOptimizer(\n",
    "    attacked_model.parameters(),\n",
    "    base_optimizer=base_optimizer,\n",
    "    scale_strength=3,  # Potenziato\n",
    "    perturb_strength=3,  # Potenziato\n",
    "    attack_type='combined'\n",
    ")\n",
    "attacked_accuracies = train_and_evaluate(attacked_model, drunk_optimizer, train_loader, test_loader, \"attacked\", num_epochs)\n",
    "\n",
    "print(\"Training the model with defense techniques...\")\n",
    "defended_model = utils_a.CNN(num_classes=4).to(device)\n",
    "defended_model.load_state_dict(state_dict)\n",
    "defended_accuracies = train_and_evaluate(defended_model, drunk_optimizer, train_loader, test_loader, \"defended\", num_epochs, adv_training=True)\n",
    "\n",
    "# Plotting all accuracies together\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.plot(range(1, num_epochs + 1), clean_accuracies, marker='o', label='Clean Model')\n",
    "plt.plot(range(1, num_epochs + 1), attacked_accuracies, marker='o', label='Attacked Model')\n",
    "plt.plot(range(1, num_epochs + 1), defended_accuracies, marker='o', label='Defended Model')\n",
    "plt.title('Training Accuracy Comparison')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy (%)')\n",
    "plt.legend()\n",
    "plt.show()"
   ],
   "id": "f0200a18fc7f5904",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training the clean model...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 71.94%\n",
      "Epoch 2/5, Accuracy: 80.02%\n",
      "Epoch 3/5, Accuracy: 81.71%\n",
      "Epoch 4/5, Accuracy: 83.81%\n",
      "Epoch 5/5, Accuracy: 84.42%\n",
      "Testing started on clean.\n",
      "Testing completed on clean.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.90      0.81      0.86       300\n",
      "  meningioma       0.77      0.86      0.81       306\n",
      "     notumor       0.95      0.97      0.96       405\n",
      "   pituitary       0.95      0.90      0.92       300\n",
      "\n",
      "    accuracy                           0.89      1311\n",
      "   macro avg       0.89      0.89      0.89      1311\n",
      "weighted avg       0.90      0.89      0.89      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.25%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model with defense techniques...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 56.30%\n",
      "Epoch 2/5, Accuracy: 56.79%\n",
      "Epoch 3/5, Accuracy: 57.46%\n",
      "Epoch 4/5, Accuracy: 57.20%\n",
      "Epoch 5/5, Accuracy: 56.14%\n",
      "Testing started on defended.\n",
      "Testing completed on defended.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.99      0.87      0.93       300\n",
      "  meningioma       0.88      0.95      0.92       306\n",
      "     notumor       0.97      1.00      0.99       405\n",
      "   pituitary       0.99      0.99      0.99       300\n",
      "\n",
      "    accuracy                           0.96      1311\n",
      "   macro avg       0.96      0.95      0.95      1311\n",
      "weighted avg       0.96      0.96      0.96      1311\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 6
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T11:24:20.368385Z",
     "start_time": "2024-08-28T09:26:29.398393Z"
    }
   },
   "cell_type": "code",
   "source": [
    "accuracy_results = {}\n",
    "\n",
    "for i in range(1, 10):\n",
    "    scale_strength = i\n",
    "    print(f\"Training the model attacked with DrunkOptimizer with scale strength = {scale_strength}...\")\n",
    "\n",
    "    # Train the attacked model\n",
    "    attacked_model = utils_a.CNN(num_classes=4).to(device)\n",
    "    attacked_model.load_state_dict(state_dict)\n",
    "    base_optimizer = optim.Adam(attacked_model.parameters(), lr=0.0001, weight_decay=1e-5)\n",
    "    drunk_optimizer = DrunkOptimizer(\n",
    "        attacked_model.parameters(),\n",
    "        base_optimizer=base_optimizer,\n",
    "        scale_strength=scale_strength,  # Potenziato\n",
    "        perturb_strength=scale_strength,  # Potenziato\n",
    "        attack_type='combined'\n",
    "    )\n",
    "    attacked_accuracies = train_and_evaluate(attacked_model, drunk_optimizer, train_loader, test_loader, \"attacked\", num_epochs)\n",
    "    # Store accuracies for plotting\n",
    "    accuracy_results[f'scale_strength_{scale_strength}'] = attacked_accuracies\n",
    "    \n",
    "    \n",
    "# Plotting all accuracies for different scale_strength values together\n",
    "plt.figure(figsize=(12, 6))\n",
    "for scale_strength, accuracies in accuracy_results.items():\n",
    "    plt.plot(range(1, num_epochs + 1), accuracies, marker='o', label=f'Scale Strength {scale_strength.split(\"_\")[-1]}')\n",
    "\n",
    "plt.title('Training Accuracy Comparison for Different Scale Strengths')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy (%)')\n",
    "plt.legend()\n",
    "plt.show()"
   ],
   "id": "db66db9c9336d113",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training the model attacked with DrunkOptimizer with scale strength = 1...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 56.37%\n",
      "Epoch 2/5, Accuracy: 55.81%\n",
      "Epoch 3/5, Accuracy: 54.50%\n",
      "Epoch 4/5, Accuracy: 55.79%\n",
      "Epoch 5/5, Accuracy: 54.87%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.99      0.93      0.96       300\n",
      "  meningioma       0.93      0.92      0.92       306\n",
      "     notumor       0.96      1.00      0.98       405\n",
      "   pituitary       0.97      0.99      0.98       300\n",
      "\n",
      "    accuracy                           0.96      1311\n",
      "   macro avg       0.96      0.96      0.96      1311\n",
      "weighted avg       0.96      0.96      0.96      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 2...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.44%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 3...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.09%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 4...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.09%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 5...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.15%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 6...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.06%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 7...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.18%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 8...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 30.20%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n",
      "Training the model attacked with DrunkOptimizer with scale strength = 9...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/torch/optim/lr_scheduler.py:60: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5, Accuracy: 29.97%\n",
      "Epoch 2/5, Accuracy: 27.92%\n",
      "Epoch 3/5, Accuracy: 27.92%\n",
      "Epoch 4/5, Accuracy: 27.92%\n",
      "Epoch 5/5, Accuracy: 27.92%\n",
      "Testing started on attacked.\n",
      "Testing completed on attacked.\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      glioma       0.00      0.00      0.00       300\n",
      "  meningioma       0.00      0.00      0.00       306\n",
      "     notumor       0.31      1.00      0.47       405\n",
      "   pituitary       0.00      0.00      0.00       300\n",
      "\n",
      "    accuracy                           0.31      1311\n",
      "   macro avg       0.08      0.25      0.12      1311\n",
      "weighted avg       0.10      0.31      0.15      1311\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
      "/Users/alessandrofolloni/PycharmProjects/break-fix/nuovo_ambiente/lib/python3.9/site-packages/sklearn/metrics/_classification.py:1531: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 1200x600 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 7
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T11:24:20.384697Z",
     "start_time": "2024-08-28T11:24:20.378780Z"
    }
   },
   "cell_type": "code",
   "source": "",
   "id": "286e9285caac4ec4",
   "outputs": [],
   "execution_count": 7
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}