mnist_example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.0.0-rc0\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras import layers\n",
    "print(tf.__version__)\n",
    "import os\n",
    "import time\n",
    "import numpy as np\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "from IPython.display import clear_output\n",
    "from model import GANomaly\n",
    "from absl import logging\n",
    "logging.set_verbosity(logging.INFO)\n",
    "logging.set_stderrthreshold(logging.INFO)\n",
    "\n",
    "def batch_resize(imgs, size: tuple):\n",
    "    img_out = np.empty((imgs.shape[0],) + size)\n",
    "    for i in range(imgs.shape[0]):\n",
    "        img_out[i] = cv2.resize(imgs[i], size, interpolation=cv2.INTER_CUBIC)\n",
    "    return img_out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Option:\n",
    "    def __init__(self):\n",
    "        self.anomaly = 2 # the anomaly digit\n",
    "        self.shuffle_buffer_size = 10000\n",
    "        self.batch_size = 300\n",
    "        self.isize = 32 # input size\n",
    "        self.ckpt_dir = \"ckpt\"\n",
    "        self.nz = 100 # latent dims\n",
    "        self.nc = 1 # input channels\n",
    "        self.ndf = 64 # number of discriminator's filters\n",
    "        self.ngf = 64 # number of generator's filters\n",
    "        self.extralayers = 0\n",
    "        self.niter = 15 # number of training epochs\n",
    "        self.lr = 2e-4 \n",
    "        self.w_adv = 1. # Adversarial loss weight\n",
    "        self.w_con = 50. # Reconstruction loss weight\n",
    "        self.w_enc = 1. # Encoder loss weight.\n",
    "        self.beta1 = 0.5\n",
    "            \n",
    "opt = Option()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 32, 32, 1) (60000, 32, 32, 1)\n"
     ]
    }
   ],
   "source": [
    "# dataset\n",
    "data_train, data_test = tf.keras.datasets.mnist.load_data()\n",
    "x_train, y_train = data_train\n",
    "x_test, y_test = data_test\n",
    "x_train = x_train.astype(np.float32)\n",
    "x_test = x_test.astype(np.float32)\n",
    "# resize to (32, 32)\n",
    "x_train = batch_resize(x_train, (32, 32))[..., None]\n",
    "x_test = batch_resize(x_test, (32, 32))[..., None]\n",
    "# normalization\n",
    "mean = x_train.mean()\n",
    "stddev = x_train.std()\n",
    "x_train = (x_train-mean)/stddev\n",
    "x_test = (x_test-mean)/stddev\n",
    "print(x_train.shape, x_train.shape)\n",
    "# define abnoraml data and normal\n",
    "# training data only contains normal\n",
    "x_train = x_train[y_train!=opt.anomaly]\n",
    "y_train = y_train[y_train!=opt.anomaly]\n",
    "y_test = (y_test==opt.anomaly).astype(np.float32)\n",
    "# tf.data.Dataset\n",
    "train_dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train))\n",
    "test_dataset = tf.data.Dataset.from_tensor_slices((x_test, y_test))\n",
    "train_dataset = train_dataset.shuffle(opt.shuffle_buffer_size).batch(opt.batch_size, drop_remainder=True)\n",
    "test_dataset = test_dataset.batch(opt.batch_size, drop_remainder=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "ganomaly = GANomaly(opt, train_dataset, valid_dataset=None, test_dataset=test_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Layer net_g is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "WARNING:tensorflow:Layer net_d is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 0, G_losses: 23.2235, D_losses: 0.6404, samples/sec: 3515.1968\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:*** \t Testing:   roc_auc: 0.4286   average_precision: 0.0818 ***\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 1, G_losses: 20.7901, D_losses: 0.0311, samples/sec: 6037.5740\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:*** \t Testing:   roc_auc: 0.5962   average_precision: 0.1223 ***\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 2, G_losses: 17.0200, D_losses: 0.0088, samples/sec: 6004.8680\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:*** \t Testing:   roc_auc: 0.9255   average_precision: 0.5721 ***\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 3, G_losses: 15.2609, D_losses: 0.0064, samples/sec: 6099.2286\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.8614   average_precision: 0.4135\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 4, G_losses: 14.3134, D_losses: 0.0053, samples/sec: 6105.4841\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:*** \t Testing:   roc_auc: 0.9431   average_precision: 0.6084 ***\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 5, G_losses: 13.7432, D_losses: 0.0031, samples/sec: 6070.9288\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9339   average_precision: 0.5572\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 6, G_losses: 13.3667, D_losses: 0.0020, samples/sec: 6191.3731\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9391   average_precision: 0.5743\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 7, G_losses: 13.1030, D_losses: 0.0026, samples/sec: 6311.5772\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9353   average_precision: 0.5508\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 8, G_losses: 12.9052, D_losses: 0.0011, samples/sec: 6319.5554\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:*** \t Testing:   roc_auc: 0.9472   average_precision: 0.5901 ***\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 9, G_losses: 12.7763, D_losses: 0.0011, samples/sec: 6096.7652\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:*** \t Testing:   roc_auc: 0.9473   average_precision: 0.5810 ***\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 10, G_losses: 12.6430, D_losses: 0.0008, samples/sec: 6177.2143\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9386   average_precision: 0.5529\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 11, G_losses: 14.9309, D_losses: 7.1135, samples/sec: 6365.5426\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9133   average_precision: 0.4745\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 12, G_losses: 15.0505, D_losses: 7.7125, samples/sec: 6409.6335\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.8933   average_precision: 0.4350\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 13, G_losses: 14.9569, D_losses: 7.7125, samples/sec: 6515.4305\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9265   average_precision: 0.5081\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=180), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:epoch: 14, G_losses: 14.8712, D_losses: 7.7125, samples/sec: 6354.1383\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:\t Testing:   roc_auc: 0.9097   average_precision: 0.4758\n"
     ]
    }
   ],
   "source": [
    "ganomaly.fit(opt.niter)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ganomaly.evaluate_best(test_dataset)"
   ]
  },
  {
   "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": 4
}