analyze_adversarial_example_detection.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "import torch\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.gridspec as gridspec\n",
    "import seaborn as sns\n",
    "sns.set_theme()\n",
    "\n",
    "from source.constants import RESULTS_PATH, PLOTS_PATH\n",
    "from source.utils.uncertainty_measures import calculate_uncertainties\n",
    "from source.utils.metrics import auroc, aupr, fpr_at_tpr_x\n",
    "\n",
    "os.makedirs(PLOTS_PATH, exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fgsm deep_ensemble resnet18 auroc\n"
     ]
    }
   ],
   "source": [
    "seed = 42\n",
    "\n",
    "methods = [\"deep_ensemble\", \"laplace\", \"mc_dropout\"]\n",
    "dataset_names = [\"cifar10\", \"cifar100\", \"svhn\", \"tin\"]\n",
    "adv_atks = [\"fgsm\", \"linfpgd\"] \n",
    "n_classes = [10, 100, 10, 200]\n",
    "models = [\"resnet18\", \"densenet169\"]\n",
    "\n",
    "method = methods[0]                         # select method \n",
    "metric = [auroc, aupr, fpr_at_tpr_x][0]     # select metric\n",
    "adv_atk = adv_atks[0]                       # select adversarial attack\n",
    "\n",
    "allow_flip = False                           # allow flipping of label for AUROC\n",
    "\n",
    "p_drop = 0.2                                # dropout rate of MC dropout     \n",
    "\n",
    "model = models[0]                           # select model\n",
    "runs = 5                                    # number of runs\n",
    "\n",
    "print(adv_atk, method, model, metric.__name__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10000, 5, 10, 10])\n",
      "torch.Size([10000, 5, 10, 100])\n",
      "torch.Size([26032, 5, 10, 10])\n",
      "torch.Size([10000, 5, 10, 200])\n",
      "torch.Size([10000, 5, 10, 10])\n",
      "torch.Size([10000, 5, 10, 100])\n",
      "torch.Size([26032, 5, 10, 10])\n",
      "torch.Size([10000, 5, 10, 200])\n"
     ]
    }
   ],
   "source": [
    "probits = list()\n",
    "y_tests = list()\n",
    "\n",
    "for dataset_name, n_class in zip(dataset_names, n_classes):\n",
    "\n",
    "    if method == \"deep_ensemble\":\n",
    "        path = os.path.join(RESULTS_PATH, f\"{dataset_name}_{model}_seed{seed}\")\n",
    "    elif method == \"laplace\":\n",
    "        path = os.path.join(RESULTS_PATH, f\"{dataset_name}_{model}_seed{seed}_laplace\")\n",
    "    elif method == \"mc_dropout\":\n",
    "        path = os.path.join(RESULTS_PATH, f\"{dataset_name}_{model}_dropout{p_drop}_seed{seed}\")\n",
    "\n",
    "    p = torch.load(os.path.join(path, f\"{dataset_name}_probits.pt\")).to(torch.float32)\n",
    "    p = p.reshape(len(p), runs, -1, n_class)\n",
    "    y_test = torch.load(os.path.join(path, f\"{dataset_name}_ys.pt\"))\n",
    "    y_tests.append(y_test)\n",
    "    probits.append(p)\n",
    "    print(probits[-1].shape)\n",
    "\n",
    "uncertainties = list()\n",
    "\n",
    "for d, ds_probits in enumerate(probits):\n",
    "    # reversed_probits = torch.flip(ds_probits, [2])\n",
    "    uncertainties.append(calculate_uncertainties(ds_probits, ds_probits))\n",
    "\n",
    "### Adversarial examples\n",
    "\n",
    "ae_probits = list()\n",
    "\n",
    "for dataset_name, n_class in zip(dataset_names, n_classes):\n",
    "\n",
    "    if method == \"deep_ensemble\":\n",
    "        path = os.path.join(RESULTS_PATH, f\"{dataset_name}_{model}_seed{seed}\")\n",
    "    elif method == \"laplace\":\n",
    "        path = os.path.join(RESULTS_PATH, f\"{dataset_name}_{model}_seed{seed}_laplace\")\n",
    "    elif method == \"mc_dropout\":\n",
    "        path = os.path.join(RESULTS_PATH, f\"{dataset_name}_{model}_dropout{p_drop}_seed{seed}\")\n",
    "\n",
    "    ae_path = os.path.join(path, \"adversarial_examples\")\n",
    "    p = torch.load(os.path.join(ae_path, f\"{adv_atk}_probits.pt\")).to(torch.float32)\n",
    "    p = p.reshape(len(p), runs, -1, n_class)\n",
    "    ae_probits.append(p)\n",
    "    print(ae_probits[-1].shape)\n",
    "\n",
    "ae_uncertainties = list()\n",
    "\n",
    "for d, ds_probits in enumerate(ae_probits):\n",
    "    # reversed_probits = torch.flip(ds_probits, [2])\n",
    "    ae_uncertainties.append(calculate_uncertainties(ds_probits, ds_probits))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cifar10 ['58.77%', '60.59%', '60.01%', '60.21%', '62.22%']\n",
      "cifar100 ['67.51%', '68.21%', '67.13%', '67.28%', '67.65%']\n",
      "svhn ['42.14%', '39.01%', '40.57%', '41.55%', '42.31%']\n",
      "tin ['52.97%', '52.55%', '52.73%', '53.39%', '52.35%']\n",
      "torch.Size([4, 26, 5])\n"
     ]
    }
   ],
   "source": [
    "unc_names = [\"A1\", \"A2\", \"A3\", \"B1\", \"B2\", \"B3\", \"C1\", \"C2\", \"C3\"]\n",
    "components = [\"TU\", \"AU\", \"EU\"]\n",
    "\n",
    "verbose = False\n",
    "\n",
    "all_perfs = list()\n",
    "\n",
    "for i in range(len(dataset_names)):\n",
    "    for c, comp in enumerate(components):\n",
    "        for measure in unc_names:\n",
    "            if comp == \"EU\" and measure == \"B2\":\n",
    "                continue # B2 is not defined for epistemic uncertainty\n",
    "            scores_clean = uncertainties[i][measure][c]\n",
    "            scores_adv = ae_uncertainties[i][measure][c]\n",
    "\n",
    "            fraction = list()\n",
    "\n",
    "            for r in range(runs):\n",
    "                \n",
    "                # mask where clean predictions of attacked model are correct\n",
    "                mask = torch.argmax(probits[i][:, r, 0], dim=-1) == y_tests[i]\n",
    "                if verbose and c == 0 and measure == \"A1\": print(dataset_names[i], \", Correctness:\", mask.float().mean())\n",
    "                # mask where adversarial predictions of attacked model are incorrect\n",
    "                mask_adv = torch.argmax(ae_probits[i][:, r, 0], dim=-1) != y_tests[i]\n",
    "                if verbose and c == 0 and measure == \"A1\": print(dataset_names[i], \", Attack Success:\", mask_adv.float().mean())\n",
    "                # mask where clean predictions of attacked model are correct and adversarial predictions are incorrect\n",
    "                mask = mask & mask_adv\n",
    "                if verbose and c == 0 and measure == \"A1\": print(dataset_names[i], \", Subset:\", mask.float().mean())\n",
    "\n",
    "                fraction.append(f\"{mask.float().mean().item() * 100:.2f}%\")\n",
    "\n",
    "                scores_clean_ = scores_clean[mask, r]\n",
    "                scores_adv_ = scores_adv[mask, r] \n",
    "\n",
    "                scores = torch.cat([scores_clean_, scores_adv_])\n",
    "                ys = torch.cat([torch.zeros(size=(len(scores_clean_), ), dtype=torch.long), \n",
    "                                torch.ones(size=(len(scores_adv_), ), dtype=torch.long)])\n",
    "                \n",
    "                # flip labels if clean model is more uncertain than attacked model\n",
    "                if allow_flip and scores_clean.mean() > scores_adv.mean():\n",
    "                    ys = 1 - ys\n",
    "\n",
    "                all_perfs.append(metric(scores, ys).item() * 100)\n",
    "            \n",
    "            if c==0 and measure == \"A1\":\n",
    "                print(dataset_names[i], fraction)\n",
    "\n",
    "all_perfs = torch.tensor(all_perfs).reshape(len(dataset_names), -1, runs)\n",
    "print(all_perfs.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot(data, title, name:str = None):\n",
    "\n",
    "    tu_colors = [\"#8FB4EF\", \"#5D8CD8\", \"#4C72B0\", \"#F2AD85\", \"#DD8452\", \"#CC6333\"]\n",
    "    au_colors = [\"#5D8CD8\", \"#DD8452\", \"#55A868\"]\n",
    "    eu_colors = [\"#8FB4EF\", \"#5D8CD8\", \"#4C72B0\", \"#F2AD85\", \"#CC6333\", \"#85BF91\", \"#55A868\", \"#418050\"]\n",
    "\n",
    "    # tu_colors = au_colors = eu_colors = sns.color_palette(\"deep\")\n",
    "\n",
    "\n",
    "    tu_labels = [\"A1\", \"A2\", \"A3\", \"B/C1\", \"B/C2\", \"B/C3\"]\n",
    "    au_labels = [\"A\", \"B\", \"C\"]\n",
    "    eu_labels = [\"A1\", \"A2\", \"A3\", \"B1\", \"B3\", \"C1\", \"C2\", \"C3\"]\n",
    "\n",
    "    # Create a figure\n",
    "    fig, axs = plt.figure(figsize=(8, 3.2)), list()\n",
    "\n",
    "    # Create a GridSpec with 1 row and 3 columns, and specify relative widths\n",
    "    gs = gridspec.GridSpec(1, 3, width_ratios=[6.25, 3.25, 8.25])\n",
    "\n",
    "    # Create subplots in the specified grid locations\n",
    "    axs.append(fig.add_subplot(gs[0]))\n",
    "    axs.append(fig.add_subplot(gs[1], sharey=axs[-1]))\n",
    "    axs.append(fig.add_subplot(gs[2], sharey=axs[-1]))\n",
    "\n",
    "    for i in range(len(tu_labels)):\n",
    "        b = axs[0].bar(tu_labels[i], data[i].mean(), yerr=data[i].std(), color=tu_colors[i], label=tu_labels[i], linewidth=0)\n",
    "        # if i >= 3:\n",
    "        #     axs[0].bar(i + b[0].get_width() / 4, data[i].mean(), color=eu_colors[2 + i], linewidth=0, width=b[0].get_width() / 2)\n",
    "    axs[0].set_xlabel(\"TU\")\n",
    "\n",
    "    for i in range(len(au_labels)):\n",
    "        tl = 9\n",
    "        axs[1].bar(au_labels[i], data[tl + i * 3].mean(), yerr=data[tl + i * 3].std(), color=au_colors[i], label=au_labels[i], linewidth=0)\n",
    "    axs[1].set_xlabel(\"AU\")\n",
    "\n",
    "    for i in range(len(eu_labels)):\n",
    "        tl = 18\n",
    "        axs[2].bar(eu_labels[i], data[tl + i].mean(), yerr=data[tl + i].std(), color=eu_colors[i], label=eu_labels[i], linewidth=0)\n",
    "    axs[2].set_xlabel(\"EU\")\n",
    "\n",
    "    # set ylim to minimum and maximum values\n",
    "    for i in range(3):\n",
    "        axs[i].hlines(data.mean(dim=-1).min() if metric == fpr_at_tpr_x else data.mean(dim=-1).max(), \n",
    "                      -0.625, len([tu_labels, au_labels, eu_labels][i]) - 0.375, colors=\"black\", linestyles=\"dashed\", alpha=0.3)\n",
    "    \n",
    "    data_range = (data.mean(dim=-1) + data.std(dim=-1)).max().item() - (data.mean(dim=-1) - data.std(dim=-1)).min().item()\n",
    "    plt.ylim([(data.mean(dim=-1) - data.std(dim=-1)).min().item() - data_range * 0.05, \n",
    "              (data.mean(dim=-1) + data.std(dim=-1)).max().item() + data_range * 0.05])\n",
    "    axs[0].set_ylabel({auroc: \"AUROC\", aupr: \"AUPR\", fpr_at_tpr_x: \"FPR95\"}[metric])\n",
    "    axs[1].set_title(title, fontsize=14)\n",
    "\n",
    "    # Optionally, hide y-axis labels for all but the first plot\n",
    "    plt.setp(axs[1].get_yticklabels(), visible=False)\n",
    "    plt.setp(axs[2].get_yticklabels(), visible=False)\n",
    "\n",
    "    for ax, n_data in zip(axs, [len(tu_labels), len(au_labels), len(eu_labels)]):\n",
    "        ax.set_xlim(-0.625, n_data - 0.375)\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n",
    "    if name is not None:\n",
    "        fig.savefig(os.path.join(PLOTS_PATH, f\"{name}.pdf\"))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x320 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt_data = torch.mean(all_perfs, dim=0)\n",
    "\n",
    "plot(plt_data, \"Overall\", f\"ae_{method}_{model}_{adv_atk}_{metric.__name__}_flip{allow_flip}_overall\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x320 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x320 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x320 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x320 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(len(dataset_names)):\n",
    "    plot(all_perfs[i], f\"{dataset_names[i].upper()}\")#, f\"{method}_{model}_{adv_atk}_{metric.__name__}_{dataset_names[i]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "quam",
   "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": "3.10.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}