codes/conv_autoencoder_lecture_2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pip install tensorflow\n",
    "pip install keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import Input, Conv2D, MaxPooling2D, UpSampling2D\n",
    "from keras.models import Model\n",
    "from keras.datasets import mnist\n",
    "from keras.callbacks import TensorBoard\n",
    "from keras import backend as K\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of encoded (None, 4, 4, 8)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "input_img = Input(shape=(28, 28, 1))    # adapt this if using 'channels_first' image data format\n",
    "\n",
    "x = Conv2D(16, (3, 3), activation='relu', padding='same')(input_img)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(8, (3, 3), activation='relu', padding='same')(x)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(8, (3, 3), activation='relu', padding='same')(x)\n",
    "encoded = MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "print(\"shape of encoded\", K.int_shape(encoded))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = Conv2D(8, (3, 3), activation='relu', padding='same')(encoded)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(8, (3, 3), activation='relu', padding='same')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(16, (3, 3), activation='relu')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "decoded = Conv2D(1, (3, 3), activation='sigmoid', padding='same')(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "autoencoder = Model(input_img, decoded)\n",
    "autoencoder.compile(optimizer='adadelta', loss='binary_crossentropy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.datasets import mnist\n",
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
    "\n",
    "x_train = x_train.astype('float32') / 255.\n",
    "x_test = x_test.astype('float32') / 255.\n",
    "x_train = np.reshape(x_train, (len(x_train), 28, 28, 1))    \n",
    "x_test = np.reshape(x_test, (len(x_test), 28, 28, 1))  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples, validate on 10000 samples\n",
      "Epoch 1/50\n",
      " - 73s - loss: 0.2182 - val_loss: 0.1676\n",
      "Epoch 2/50\n",
      " - 78s - loss: 0.1552 - val_loss: 0.1447\n",
      "Epoch 3/50\n",
      " - 63s - loss: 0.1419 - val_loss: 0.1381\n",
      "Epoch 4/50\n",
      " - 63s - loss: 0.1336 - val_loss: 0.1287\n",
      "Epoch 5/50\n",
      " - 63s - loss: 0.1287 - val_loss: 0.1244\n",
      "Epoch 6/50\n",
      " - 63s - loss: 0.1252 - val_loss: 0.1245\n",
      "Epoch 7/50\n",
      " - 63s - loss: 0.1224 - val_loss: 0.1171\n",
      "Epoch 8/50\n",
      " - 63s - loss: 0.1204 - val_loss: 0.1209\n",
      "Epoch 9/50\n",
      " - 64s - loss: 0.1190 - val_loss: 0.1208\n",
      "Epoch 10/50\n",
      " - 63s - loss: 0.1173 - val_loss: 0.1199\n",
      "Epoch 11/50\n",
      " - 63s - loss: 0.1160 - val_loss: 0.1126\n",
      "Epoch 12/50\n",
      " - 64s - loss: 0.1146 - val_loss: 0.1110\n",
      "Epoch 13/50\n",
      " - 64s - loss: 0.1135 - val_loss: 0.1130\n",
      "Epoch 14/50\n",
      " - 64s - loss: 0.1131 - val_loss: 0.1113\n",
      "Epoch 15/50\n",
      " - 64s - loss: 0.1121 - val_loss: 0.1109\n",
      "Epoch 16/50\n",
      " - 64s - loss: 0.1117 - val_loss: 0.1093\n",
      "Epoch 17/50\n",
      " - 64s - loss: 0.1108 - val_loss: 0.1091\n",
      "Epoch 18/50\n",
      " - 64s - loss: 0.1101 - val_loss: 0.1089\n",
      "Epoch 19/50\n",
      " - 64s - loss: 0.1095 - val_loss: 0.1101\n",
      "Epoch 20/50\n",
      " - 64s - loss: 0.1089 - val_loss: 0.1093\n",
      "Epoch 21/50\n",
      " - 64s - loss: 0.1083 - val_loss: 0.1083\n",
      "Epoch 22/50\n",
      " - 63s - loss: 0.1079 - val_loss: 0.1076\n",
      "Epoch 23/50\n",
      " - 63s - loss: 0.1072 - val_loss: 0.1069\n",
      "Epoch 24/50\n",
      " - 65s - loss: 0.1066 - val_loss: 0.1055\n",
      "Epoch 25/50\n",
      " - 64s - loss: 0.1061 - val_loss: 0.1059\n",
      "Epoch 26/50\n",
      " - 64s - loss: 0.1056 - val_loss: 0.1060\n",
      "Epoch 27/50\n",
      " - 65s - loss: 0.1055 - val_loss: 0.1042\n",
      "Epoch 28/50\n",
      " - 65s - loss: 0.1052 - val_loss: 0.1031\n",
      "Epoch 29/50\n",
      " - 65s - loss: 0.1050 - val_loss: 0.1041\n",
      "Epoch 30/50\n",
      " - 66s - loss: 0.1047 - val_loss: 0.1034\n",
      "Epoch 31/50\n",
      " - 65s - loss: 0.1045 - val_loss: 0.1027\n",
      "Epoch 32/50\n",
      " - 65s - loss: 0.1040 - val_loss: 0.1028\n",
      "Epoch 33/50\n",
      " - 65s - loss: 0.1040 - val_loss: 0.1020\n",
      "Epoch 34/50\n",
      " - 65s - loss: 0.1037 - val_loss: 0.1032\n",
      "Epoch 35/50\n",
      " - 65s - loss: 0.1034 - val_loss: 0.1031\n",
      "Epoch 36/50\n",
      " - 66s - loss: 0.1031 - val_loss: 0.1002\n",
      "Epoch 37/50\n",
      " - 65s - loss: 0.1029 - val_loss: 0.1005\n",
      "Epoch 38/50\n",
      " - 64s - loss: 0.1028 - val_loss: 0.1038\n",
      "Epoch 39/50\n",
      " - 65s - loss: 0.1026 - val_loss: 0.0990\n",
      "Epoch 40/50\n",
      " - 65s - loss: 0.1024 - val_loss: 0.1037\n",
      "Epoch 41/50\n",
      " - 63s - loss: 0.1022 - val_loss: 0.1028\n",
      "Epoch 42/50\n",
      " - 65s - loss: 0.1021 - val_loss: 0.1006\n",
      "Epoch 43/50\n",
      " - 65s - loss: 0.1020 - val_loss: 0.1014\n",
      "Epoch 44/50\n",
      " - 65s - loss: 0.1016 - val_loss: 0.1006\n",
      "Epoch 45/50\n",
      " - 65s - loss: 0.1016 - val_loss: 0.1013\n",
      "Epoch 46/50\n",
      " - 66s - loss: 0.1013 - val_loss: 0.0994\n",
      "Epoch 47/50\n",
      " - 65s - loss: 0.1010 - val_loss: 0.1013\n",
      "Epoch 48/50\n",
      " - 65s - loss: 0.1011 - val_loss: 0.1010\n",
      "Epoch 49/50\n",
      " - 65s - loss: 0.1009 - val_loss: 0.0987\n",
      "Epoch 50/50\n",
      " - 65s - loss: 0.1004 - val_loss: 0.1000\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x11f9ab4e0>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "autoencoder.fit(x_train, x_train, epochs=50, batch_size=128, shuffle=True, validation_data=(x_test, x_test),\n",
    "                callbacks=[TensorBoard(log_dir='conv_autoencoder')], verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "decoded_imgs = autoencoder.predict(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x400 with 20 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = 10\n",
    "plt.figure(figsize=(10, 4), dpi=100)\n",
    "for i in range(n):\n",
    "    # display original\n",
    "    ax = plt.subplot(2, n, i + 1)\n",
    "    plt.imshow(x_test[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.set_axis_off()\n",
    "\n",
    "    # display reconstruction\n",
    "    ax = plt.subplot(2, n, i + n + 1)\n",
    "    plt.imshow(decoded_imgs[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.set_axis_off()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "encoder = Model(input_img, encoded)\n",
    "encoded_imgs = encoder.predict(x_test)\n",
    "\n",
    "# save latent space features 128-d vector\n",
    "pickle.dump(encoded_imgs, open('conv_autoe_features.pickle', 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x400 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = 10\n",
    "plt.figure(figsize=(10, 4), dpi=100)\n",
    "for i in range(n):\n",
    "    ax = plt.subplot(1, n, i + 1)\n",
    "    plt.imshow(encoded_imgs[i].reshape(4, 4 * 8).T)\n",
    "    plt.gray()\n",
    "    ax.set_axis_off()\n",
    "\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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}