Figure 2 - GenSample.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import scipy\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn import mixture, preprocessing, datasets\n",
    "\n",
    "from importlib import reload\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import patches\n",
    "\n",
    "import torch.utils.data as data_utils\n",
    "\n",
    "import utils.models as models\n",
    "import utils.plotting as plotting\n",
    "import utils.dataloaders as dl\n",
    "import utils.traintest as tt\n",
    "import utils.adversarial as adv\n",
    "import utils.eval as ev\n",
    "import model_params as params\n",
    "import utils.resnet_orig as resnet\n",
    "import utils.gmm_helpers as gmm_helpers\n",
    "\n",
    "import model_paths"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = 'MNIST'\n",
    "\n",
    "file_dict = {'MNIST' :'samples_steps500_alpha3.0_restarts50_batches2_batch_size100_MNIST_2019-11-08 08:46:38.902661_MNIST.pth',\n",
    "             'FMNIST':'samples_steps500_alpha3.0_restarts50_batches2_batch_size100_FMNIST_2019-11-08 12:10:48.877278_FMNIST.pth',\n",
    "             'SVHN' : 'samples_steps500_alpha3.0_restarts50_batches2_batch_size100_SVHN_2019-11-08 14:45:10.708336_SVHN.pth',\n",
    "             'CIFAR10':'samples_steps500_alpha3.0_restarts50_batches2_batch_size100_CIFAR10_2019-11-08 08:46:55.962307_CIFAR10.pth',\n",
    "             'CIFAR100':'samples_steps500_alpha3.0_restarts50_batches2_batch_size100_CIFAR100_2019-11-10 17:12:27.811439_CIFAR100.pth',\n",
    "            }\n",
    "\n",
    "file = file_dict[dataset]\n",
    "\n",
    "batch = torch.load('results/backup/' + file)\n",
    "\n",
    "model_params = params.params_dict[dataset]()\n",
    "model_path = model_paths.model_dict[dataset]() \n",
    "model_list = [torch.load(file).cpu().eval() for file in list(model_path.file_dict.values())[:-1]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device('cuda:4')\n",
    "\n",
    "num_of_models = len(model_list)\n",
    "\n",
    "num = batch.samples.shape[0]*batch.samples.shape[2]\n",
    "\n",
    "if dataset in ['MNIST', 'FMNIST']:\n",
    "    noise = batch.samples.transpose(0,1).contiguous().view(num_of_models, num, 1, 28, 28)\n",
    "elif dataset in ['CIFAR10', 'SVHN', 'CIFAR100']:\n",
    "    noise = batch.samples.transpose(0,1).contiguous().view(num_of_models, num, 3, 32, 32)\n",
    "    \n",
    "noise_list = []\n",
    "for i in range(num_of_models):\n",
    "    noise_list.append(noise[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y = [model(no).max(1) for (no, model) in zip(noise_list, model_list)]\n",
    "conf = [y[0][0].exp().item() for y in Y]\n",
    "pred = [y[1][0].item() for y in Y]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "medians = [tt.get_median(model.to(device), device, model_params.test_loader).item()\n",
    "           for model in model_list]\n",
    "model_list = [model.cpu() for model in model_list]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([170])"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conf[7][0:170].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12\n"
     ]
    }
   ],
   "source": [
    "conf = torch.stack([y[0] for y in Y], 0)\n",
    "index = conf.exp().sum(0).argmax() \n",
    "index = 12\n",
    "#index = conf.exp()[7][0:170].argmax() \n",
    "print(index)\n",
    "#index = 10\n",
    "\n",
    "if dataset in ['MNIST', 'FMNIST']:\n",
    "    seed = batch.seeds.view(-1,1,28,28)[index]\n",
    "elif dataset in ['CIFAR10', 'SVHN', 'CIFAR100']:\n",
    "    seed = batch.seeds.view(-1,3,32,32)[index]\n",
    "\n",
    "adv_noise = noise[:,index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_names = list(model_path.file_dict.keys())[:-1]\n",
    "\n",
    "plt.subplot(1, num_of_models+1, 1)\n",
    "#plt.ylabel(' ' + dataset + '\\n', rotation=90, horizontalalignment='left', va='center',\n",
    "#          fontsize=14, position=(0,0))\n",
    "\n",
    "if dataset in ['MNIST', 'FMNIST']:\n",
    "    plt.imshow(seed.squeeze().detach().cpu(), cmap='gray', interpolation='none')\n",
    "elif dataset in ['CIFAR10', 'SVHN', 'CIFAR100']:\n",
    "    plt.imshow(seed.transpose(0,2).transpose(0,1).detach().cpu(), interpolation='none')\n",
    "\n",
    "plt.xticks([])\n",
    "plt.yticks([])\n",
    "\n",
    "for i in range(num_of_models):\n",
    "    plt.subplot(1, num_of_models + 1, 2 + i)\n",
    "    #if i==0:\n",
    "        #plt.ylabel(' ' + dataset + '\\n', rotation=90, horizontalalignment='left', va='center',\n",
    "        #           fontsize=14, position=(0,0))\n",
    "        \n",
    "    if dataset in ['MNIST', 'SVHN']:\n",
    "        string = model_names[i] + '\\n'\n",
    "    else:\n",
    "        string = ''\n",
    "\n",
    "    subtitle_string = ( str(plotting.classes_dict[dataset][pred[i]] )\n",
    "                       + \": %.1f\" % conf[i][index].exp().item() )\n",
    "    \n",
    "    title_string = model_names[i]\n",
    "    \n",
    "    if dataset=='MNIST':\n",
    "        plt.title(title_string, fontsize=7)\n",
    "    #plt.title(string)\n",
    "    plt.xlabel(subtitle_string, fontsize=5)\n",
    "\n",
    "    if dataset in ['MNIST', 'FMNIST']:\n",
    "        plt.imshow(adv_noise[i].squeeze().detach().cpu(), cmap='gray', interpolation='none')\n",
    "    elif dataset in ['CIFAR10', 'SVHN', 'CIFAR100']:\n",
    "        plt.imshow(adv_noise[i].transpose(0,2).transpose(0,1).detach().cpu(), interpolation='none')\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "    \n",
    "# plt.tight_layout()\n",
    "plt.subplots_adjust(left=None, bottom=None, right=None, top=None, wspace=.16, hspace=None)\n",
    "myplot = plt.gcf()\n",
    "\n",
    "#myplot.savefig('results/' + dataset + '_sample_all_models' + '.eps', format='eps', \n",
    "#               bbox_inches = 'tight', pad_inches = .01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_names = list(model_path.file_dict.keys())\n",
    "model_names =  ['Base', 'MCD', 'GAN', 'ODIN', 'Maha', 'ACET', 'OE', 'CCU', 'CCUs']\n",
    "model_names = ['Base', 'MCD', 'EDL', 'DE', 'GAN', 'ODIN', \n",
    "               'Maha', 'ACET', 'ACET2', 'OE', 'CCU', 'CCUs']\n",
    "\n",
    "fontsize = 28\n",
    "fontsize_title = 34\n",
    "pad = 5\n",
    "model_order = [0, 1, 3, 6, 4, 2, 8, 5, 7]\n",
    "model_order = list(np.arange(11))\n",
    "\n",
    "master_list = [[]] + [[x] for x in list(range(11))]\n",
    "\n",
    "for model_order in master_list:\n",
    "    num_of_models = len(model_order)\n",
    "\n",
    "    if num_of_models==0:\n",
    "\n",
    "        plt.subplot(1, 1, 1)\n",
    "        if dataset in ['MNIST']:\n",
    "            plt.title('Seed', fontsize=fontsize_title, pad=pad)\n",
    "        #plt.ylabel(' ' + dataset + '\\n', rotation=90, horizontalalignment='left', va='center',\n",
    "        #          fontsize=14, position=(0,0))\n",
    "\n",
    "        if dataset in ['MNIST', 'FMNIST']:\n",
    "            plt.imshow(seed.squeeze().detach().cpu(), cmap='gray', interpolation='none')\n",
    "        elif dataset in ['CIFAR10', 'SVHN', 'CIFAR100']:\n",
    "            plt.imshow(seed.transpose(0,2).transpose(0,1).detach().cpu(), interpolation='none')\n",
    "\n",
    "        plt.xlabel('.', fontsize=fontsize, labelpad=3, color='white')\n",
    "        plt.xticks([])\n",
    "        plt.yticks([])\n",
    "\n",
    "    for i, j in zip(model_order, range(num_of_models)):\n",
    "\n",
    "        plt.subplot(1, num_of_models, 1 + j)\n",
    "        #if i==0:\n",
    "            #plt.ylabel(' ' + dataset + '\\n', rotation=90, horizontalalignment='left', va='center',\n",
    "            #           fontsize=14, position=(0,0))\n",
    "\n",
    "        if dataset in ['MNIST']:\n",
    "            string2 = model_names[i] \n",
    "        else:\n",
    "            string2 = ''\n",
    "        \n",
    "\n",
    "        string = ( str(plotting.classes_dict[dataset][pred[i]] )\n",
    "                   + \": %.2f\" % conf[i][index].exp().item() )\n",
    "\n",
    "        if model_names[i] in ['MCD', 'Maha']:\n",
    "            string = ( str(plotting.classes_dict[dataset][pred[i]] )\n",
    "                   + \": %.2f\" % -conf[i][index].item() + '*' )\n",
    "        \n",
    "        #string2 = model_names[i] \n",
    "\n",
    "        plt.title(string2, fontsize=fontsize_title, pad=pad)\n",
    "        \n",
    "        if conf[i][index] >= medians[i] or model_names[i]=='MCD':\n",
    "            plt.xlabel(string, fontsize=fontsize, labelpad=3, \n",
    "                       color='black')\n",
    "        else:\n",
    "            plt.xlabel(string, fontsize=fontsize, labelpad=3, \n",
    "                       color='black', fontweight='bold')\n",
    "\n",
    "        if dataset in ['MNIST', 'FMNIST']:\n",
    "            plt.imshow(adv_noise[i].squeeze().detach().cpu(), cmap='gray', interpolation='none')\n",
    "        elif dataset in ['CIFAR10', 'SVHN', 'CIFAR100']:\n",
    "            plt.imshow(adv_noise[i].transpose(0,2).transpose(0,1).detach().cpu(), interpolation='none')\n",
    "        plt.xticks([])\n",
    "        plt.yticks([])\n",
    "\n",
    "\n",
    "    # plt.tight_layout()\n",
    "    plt.subplots_adjust(left=None, bottom=None, right=None, top=None, wspace=.12, hspace=.2)\n",
    "    myplot = plt.gcf()\n",
    "\n",
    "    name = model_names[model_order[0]] if num_of_models==1 else 'seed'\n",
    "    myplot.savefig('results/Samples/'+ dataset + '/' + name + '.eps', format='eps', \n",
    "                   bbox_inches = 'tight', pad_inches = .01)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}