capsule_gan.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modules imported.\n",
      "Devices: [name: \"/device:CPU:0\"\n",
      "device_type: \"CPU\"\n",
      "memory_limit: 268435456\n",
      "locality {\n",
      "}\n",
      "incarnation: 9011255001604530858\n",
      ", name: \"/device:GPU:0\"\n",
      "device_type: \"GPU\"\n",
      "memory_limit: 136396800\n",
      "locality {\n",
      "  bus_id: 1\n",
      "}\n",
      "incarnation: 16670803991041366025\n",
      "physical_device_desc: \"device: 0, name: GeForce GTX 660, pci bus id: 0000:01:00.0, compute capability: 3.0\"\n",
      "]\n",
      "Default GPU Device: /device:GPU:0\n"
     ]
    }
   ],
   "source": [
    "# math libraries\n",
    "import numpy as np\n",
    "\n",
    "# ml libraries\n",
    "import tensorflow as tf\n",
    "from keras import layers, models, optimizers\n",
    "from keras import backend as K\n",
    "from keras.utils import to_categorical\n",
    "from keras.datasets import mnist, cifar10\n",
    "from keras.layers import Input, Dense, Reshape, Flatten, Dropout, Lambda, Concatenate, Multiply\n",
    "from keras.layers import BatchNormalization, Activation, ZeroPadding2D\n",
    "from keras.layers.advanced_activations import LeakyReLU\n",
    "from keras.layers.convolutional import UpSampling2D, Conv2D\n",
    "from keras.models import Sequential, Model\n",
    "from keras.optimizers import Adam\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras import callbacks\n",
    "\n",
    "# visualization\n",
    "import skimage\n",
    "from skimage import data, color, exposure\n",
    "from skimage.transform import resize\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "# sys and helpers\n",
    "import sys\n",
    "import os\n",
    "import glob\n",
    "from tqdm import tqdm\n",
    "\n",
    "print('Modules imported.')\n",
    "\n",
    "\n",
    "\n",
    "# device check\n",
    "from tensorflow.python.client import device_lib\n",
    "print('Devices:', device_lib.list_local_devices())\n",
    "\n",
    "# GPU check\n",
    "if not tf.test.gpu_device_name():\n",
    "    print('No GPU found.')\n",
    "else:\n",
    "    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Loading the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset shape: (50000, 32, 32, 3), Image shape: [32, 32, 3]\n"
     ]
    }
   ],
   "source": [
    "def load_dataset(dataset, width, height, channels):\n",
    "    \n",
    "    if dataset == 'mnist':\n",
    "        # load MNIST data\n",
    "        (X_train, y_train), (X_test, y_test) = mnist.load_data()\n",
    "\n",
    "        # rescale -1 to 1\n",
    "        X_train = (X_train.astype(np.float32) - 127.5) / 127.5\n",
    "        X_train = np.expand_dims(X_train, axis=3)\n",
    "        \n",
    "    if dataset == 'cifar10':\n",
    "        # load CIFAR10 data\n",
    "        (X_train, y_train), (X_test, y_test) = cifar10.load_data()\n",
    "\n",
    "        # rescale -1 to 1\n",
    "        X_train = (X_train.astype(np.float32) - 127.5) / 127.5\n",
    "\n",
    "        \n",
    "        \n",
    "    # defining input dims\n",
    "    img_rows = width\n",
    "    img_cols = height\n",
    "    channels = channels\n",
    "    img_shape = [img_rows, img_cols, channels]\n",
    "    \n",
    "    return X_train, img_shape\n",
    "    \n",
    "\n",
    "# if MNIST ('mnist', 28, 28, 1) if CIFAR10 ('cifar10', 32, 32, 3)\n",
    "dataset, shape = load_dataset('cifar10', 32, 32, 3)\n",
    "print('Dataset shape: {0}, Image shape: {1}'.format(dataset.shape, shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Helper functions for Capsule layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# squash function of capsule layers, borrowed from Xifeng Guo's implementation of Keras CapsNet `https://github.com/XifengGuo/CapsNet-Keras`\n",
    "def squash(vectors, axis=-1):\n",
    "    \"\"\"\n",
    "    The non-linear activation used in Capsule. It drives the length of a large vector to near 1 and small vector to 0\n",
    "    :param vectors: some vectors to be squashed, N-dim tensor\n",
    "    :param axis: the axis to squash\n",
    "    :return: a Tensor with same shape as input vectors\n",
    "    \"\"\"\n",
    "    s_squared_norm = K.sum(K.square(vectors), axis, keepdims=True)\n",
    "    scale = s_squared_norm / (1 + s_squared_norm) / K.sqrt(s_squared_norm + K.epsilon())\n",
    "    return scale * vectors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Defining the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# discriminator structure\n",
    "def build_discriminator():\n",
    "\n",
    "    \"\"\"\n",
    "    This is the part my 'Capsule Layer as a Discriminator in Generative Adversarial Networks' paper focuses on,\n",
    "    as it introduces a new structure to the discriminator of DCGAN by using Capsule Layers architecture from original\n",
    "    'Dynamic Routing Between Capsules' paper by S. Sabour, N. Frosst and G. Hinton.\n",
    "    \n",
    "    Discriminator takes real/generated images and outputs its prediction.\n",
    "    \"\"\"\n",
    "\n",
    "    # depending on dataset we define input shape for our network\n",
    "    img = Input(shape=(shape[0], shape[1], shape[2]))\n",
    "\n",
    "    # first typical convlayer outputs a 20x20x256 matrix\n",
    "    x = Conv2D(filters=256, kernel_size=9, strides=1, padding='valid', name='conv1')(img)\n",
    "    x = LeakyReLU()(x)\n",
    "    \n",
    "    # original 'Dynamic Routing Between Capsules' paper does not include the batch norm layer after the first conv group\n",
    "    x = BatchNormalization(momentum=0.8)(x)\n",
    "\n",
    "    \n",
    "    \"\"\"\n",
    "    NOTE: Capsule architecture starts from here.\n",
    "    \"\"\"\n",
    "    #\n",
    "    # primarycaps coming first\n",
    "    #\n",
    "    \n",
    "    # filters 256 (n_vectors=8 * channels=32)\n",
    "    x = Conv2D(filters=8 * 32, kernel_size=9, strides=2, padding='valid', name='primarycap_conv2')(x)\n",
    "    \n",
    "    # reshape into the 8D vector for all 32 feature maps combined\n",
    "    # (primary capsule has collections of activations which denote orientation of the digit\n",
    "    # while intensity of the vector which denotes the presence of the digit)\n",
    "    x = Reshape(target_shape=[-1, 8], name='primarycap_reshape')(x)\n",
    "    \n",
    "    # the purpose is to output a number between 0 and 1 for each capsule where the length of the input decides the amount\n",
    "    x = Lambda(squash, name='primarycap_squash')(x)\n",
    "    x = BatchNormalization(momentum=0.8)(x)\n",
    "\n",
    "\n",
    "    #\n",
    "    # digitcaps are here\n",
    "    #\n",
    "    \"\"\"\n",
    "    NOTE: My approach is a simplified version of digitcaps i.e. without expanding dimensions into\n",
    "    [None, 1, input_n_vectors, input_dim_capsule (feature maps)]\n",
    "    and tiling it into [None, num_capsule, input_n_vectors, input_dim_capsule (feature maps)].\n",
    "    Instead I replace it with ordinary Keras Dense layers as weight holders in the following lines.\n",
    "    \n",
    "    ANY CORRECTIONS ARE APPRECIATED IN THIS PART, PLEASE SUBMIT PULL REQUESTS!\n",
    "    \"\"\"\n",
    "    x = Flatten()(x)\n",
    "    # capsule (i) in a lower-level layer needs to decide how to send its output vector to higher-level capsules (j)\n",
    "    # it makes this decision by changing scalar weight (c=coupling coefficient) that will multiply its output vector and then be treated as input to a higher-level capsule\n",
    "    #\n",
    "    # uhat = prediction vector, w = weight matrix but will act as a dense layer, u = output from a previous layer\n",
    "    # uhat = u * w\n",
    "    # neurons 160 (num_capsules=10 * num_vectors=16)\n",
    "    uhat = Dense(160, kernel_initializer='he_normal', bias_initializer='zeros', name='uhat_digitcaps')(x)\n",
    "    \n",
    "    # c = coupling coefficient (softmax over the bias weights, log prior) | \"the coupling coefficients between capsule (i) and all the capsules in the layer above sum to 1\"\n",
    "    # we treat the coupling coefficiant as a softmax over bias weights from the previous dense layer\n",
    "    c = Activation('softmax', name='softmax_digitcaps1')(uhat) # softmax will make sure that each weight c_ij is a non-negative number and their sum equals to one\n",
    "    \n",
    "    # s_j (output of the current capsule level) = uhat * c\n",
    "    c = Dense(160)(c) # compute s_j\n",
    "    x = Multiply()([uhat, c])\n",
    "    \"\"\"\n",
    "    NOTE: Squashing the capsule outputs creates severe blurry artifacts, thus we replace it with Leaky ReLu.\n",
    "    \"\"\"\n",
    "    s_j = LeakyReLU()(x)\n",
    "\n",
    "\n",
    "    #\n",
    "    # we will repeat the routing part 2 more times (num_routing=3) to unfold the loop\n",
    "    #\n",
    "    c = Activation('softmax', name='softmax_digitcaps2')(s_j) # softmax will make sure that each weight c_ij is a non-negative number and their sum equals to one\n",
    "    c = Dense(160)(c) # compute s_j\n",
    "    x = Multiply()([uhat, c])\n",
    "    s_j = LeakyReLU()(x)\n",
    "\n",
    "    c = Activation('softmax', name='softmax_digitcaps3')(s_j) # softmax will make sure that each weight c_ij is a non-negative number and their sum equals to one\n",
    "    c = Dense(160)(c) # compute s_j\n",
    "    x = Multiply()([uhat, c])\n",
    "    s_j = LeakyReLU()(x)\n",
    "\n",
    "    pred = Dense(1, activation='sigmoid')(s_j)\n",
    "\n",
    "    \n",
    "    return Model(img, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DISCRIMINATOR:\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_14 (InputLayer)           (None, 32, 32, 3)    0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1 (Conv2D)                  (None, 24, 24, 256)  62464       input_14[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "leaky_re_lu_9 (LeakyReLU)       (None, 24, 24, 256)  0           conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_26 (BatchNo (None, 24, 24, 256)  1024        leaky_re_lu_9[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "primarycap_conv2 (Conv2D)       (None, 8, 8, 256)    5308672     batch_normalization_26[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "primarycap_reshape (Reshape)    (None, 2048, 8)      0           primarycap_conv2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "primarycap_squash (Lambda)      (None, 2048, 8)      0           primarycap_reshape[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_27 (BatchNo (None, 2048, 8)      32          primarycap_squash[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "flatten_3 (Flatten)             (None, 16384)        0           batch_normalization_27[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "uhat_digitcaps (Dense)          (None, 160)          2621600     flatten_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "softmax_digitcaps1 (Activation) (None, 160)          0           uhat_digitcaps[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_17 (Dense)                (None, 160)          25760       softmax_digitcaps1[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "multiply_7 (Multiply)           (None, 160)          0           uhat_digitcaps[0][0]             \n",
      "                                                                 dense_17[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "leaky_re_lu_10 (LeakyReLU)      (None, 160)          0           multiply_7[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "softmax_digitcaps2 (Activation) (None, 160)          0           leaky_re_lu_10[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_18 (Dense)                (None, 160)          25760       softmax_digitcaps2[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "multiply_8 (Multiply)           (None, 160)          0           uhat_digitcaps[0][0]             \n",
      "                                                                 dense_18[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "leaky_re_lu_11 (LeakyReLU)      (None, 160)          0           multiply_8[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "softmax_digitcaps3 (Activation) (None, 160)          0           leaky_re_lu_11[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_19 (Dense)                (None, 160)          25760       softmax_digitcaps3[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "multiply_9 (Multiply)           (None, 160)          0           uhat_digitcaps[0][0]             \n",
      "                                                                 dense_19[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "leaky_re_lu_12 (LeakyReLU)      (None, 160)          0           multiply_9[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_20 (Dense)                (None, 1)            161         leaky_re_lu_12[0][0]             \n",
      "==================================================================================================\n",
      "Total params: 8,071,233\n",
      "Trainable params: 8,070,705\n",
      "Non-trainable params: 528\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# build and compile the discriminator\n",
    "discriminator = build_discriminator()\n",
    "print('DISCRIMINATOR:')\n",
    "discriminator.summary()\n",
    "discriminator.compile(loss='binary_crossentropy', optimizer=Adam(0.0002, 0.5), metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# generator structure\n",
    "def build_generator():\n",
    "\n",
    "    \"\"\"\n",
    "    Generator follows the DCGAN architecture and creates generated image representations through learning.\n",
    "    \"\"\"\n",
    "\n",
    "    noise_shape = (100,)\n",
    "    x_noise = Input(shape=noise_shape)\n",
    "\n",
    "    # we apply different kernel sizes in order to match the original image size\n",
    "    \n",
    "    if (shape[0] == 28 and shape[1] == 28):\n",
    "        x = Dense(128 * 7 * 7, activation=\"relu\")(x_noise)\n",
    "        x = Reshape((7, 7, 128))(x)\n",
    "        x = BatchNormalization(momentum=0.8)(x)\n",
    "        x = UpSampling2D()(x)\n",
    "        x = Conv2D(128, kernel_size=3, padding=\"same\")(x)\n",
    "        x = Activation(\"relu\")(x)\n",
    "        x = BatchNormalization(momentum=0.8)(x)\n",
    "        x = UpSampling2D()(x)\n",
    "        x = Conv2D(64, kernel_size=3, padding=\"same\")(x)\n",
    "        x = Activation(\"relu\")(x)\n",
    "        x = BatchNormalization(momentum=0.8)(x)\n",
    "        x = Conv2D(1, kernel_size=3, padding=\"same\")(x)\n",
    "        gen_out = Activation(\"tanh\")(x)\n",
    "        \n",
    "        return Model(x_noise, gen_out)\n",
    "\n",
    "    if (shape[0] == 32 and shape[1] == 32):\n",
    "        x = Dense(128 * 8 * 8, activation=\"relu\")(x_noise)\n",
    "        x = Reshape((8, 8, 128))(x)\n",
    "        x = BatchNormalization(momentum=0.8)(x)\n",
    "        x = UpSampling2D()(x)\n",
    "        x = Conv2D(128, kernel_size=3, padding=\"same\")(x)\n",
    "        x = Activation(\"relu\")(x)\n",
    "        x = BatchNormalization(momentum=0.8)(x)\n",
    "        x = UpSampling2D()(x)\n",
    "        x = Conv2D(64, kernel_size=3, padding=\"same\")(x)\n",
    "        x = Activation(\"relu\")(x)\n",
    "        x = BatchNormalization(momentum=0.8)(x)\n",
    "        x = Conv2D(3, kernel_size=3, padding=\"same\")(x)\n",
    "        gen_out = Activation(\"tanh\")(x)\n",
    "\n",
    "        \n",
    "        return Model(x_noise, gen_out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GENERATOR:\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_15 (InputLayer)        (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "dense_21 (Dense)             (None, 8192)              827392    \n",
      "_________________________________________________________________\n",
      "reshape_9 (Reshape)          (None, 8, 8, 128)         0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_28 (Batc (None, 8, 8, 128)         512       \n",
      "_________________________________________________________________\n",
      "up_sampling2d_15 (UpSampling (None, 16, 16, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_22 (Conv2D)           (None, 16, 16, 128)       147584    \n",
      "_________________________________________________________________\n",
      "activation_22 (Activation)   (None, 16, 16, 128)       0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_29 (Batc (None, 16, 16, 128)       512       \n",
      "_________________________________________________________________\n",
      "up_sampling2d_16 (UpSampling (None, 32, 32, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_23 (Conv2D)           (None, 32, 32, 64)        73792     \n",
      "_________________________________________________________________\n",
      "activation_23 (Activation)   (None, 32, 32, 64)        0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_30 (Batc (None, 32, 32, 64)        256       \n",
      "_________________________________________________________________\n",
      "conv2d_24 (Conv2D)           (None, 32, 32, 3)         1731      \n",
      "_________________________________________________________________\n",
      "activation_24 (Activation)   (None, 32, 32, 3)         0         \n",
      "=================================================================\n",
      "Total params: 1,051,779\n",
      "Trainable params: 1,051,139\n",
      "Non-trainable params: 640\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# build and compile the generator\n",
    "generator = build_generator()\n",
    "print('GENERATOR:')\n",
    "generator.summary()\n",
    "generator.compile(loss='binary_crossentropy', optimizer=Adam(0.0002, 0.5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "# feeding noise to generator\n",
    "z = Input(shape=(100,))\n",
    "img = generator(z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for the combined model we will only train the generator\n",
    "discriminator.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "# try to discriminate generated images\n",
    "valid = discriminator(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "COMBINED:\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_16 (InputLayer)        (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "model_12 (Model)             (None, 32, 32, 3)         1051779   \n",
      "_________________________________________________________________\n",
      "model_11 (Model)             (None, 1)                 8071233   \n",
      "=================================================================\n",
      "Total params: 9,123,012\n",
      "Trainable params: 1,051,139\n",
      "Non-trainable params: 8,071,873\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# the combined model (stacked generator and discriminator) takes\n",
    "# noise as input => generates images => determines validity \n",
    "combined = Model(z, valid)\n",
    "print('COMBINED:')\n",
    "combined.summary()\n",
    "combined.compile(loss='binary_crossentropy', optimizer=Adam(0.0002, 0.5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# loss values for further plotting\n",
    "D_L_REAL = []\n",
    "D_L_FAKE = []\n",
    "D_L = []\n",
    "D_ACC = []\n",
    "G_L = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(dataset_title, epochs, batch_size=32, save_interval=50):\n",
    "\n",
    "        half_batch = int(batch_size / 2)\n",
    "\n",
    "        for epoch in range(epochs):\n",
    "\n",
    "            # ---------------------\n",
    "            #  Train Discriminator\n",
    "            # ---------------------\n",
    "\n",
    "            # select a random half batch of images\n",
    "            idx = np.random.randint(0, dataset.shape[0], half_batch)\n",
    "            imgs = dataset[idx]\n",
    "\n",
    "            noise = np.random.normal(0, 1, (half_batch, 100))\n",
    "\n",
    "            # generate a half batch of new images\n",
    "            gen_imgs = generator.predict(noise)\n",
    "\n",
    "            # train the discriminator by feeding both real and fake (generated) images one by one\n",
    "            d_loss_real = discriminator.train_on_batch(imgs, np.ones((half_batch, 1))*0.9) # 0.9 for label smoothing\n",
    "            d_loss_fake = discriminator.train_on_batch(gen_imgs, np.zeros((half_batch, 1)))\n",
    "            d_loss = 0.5 * np.add(d_loss_real, d_loss_fake)\n",
    "\n",
    "\n",
    "            # ---------------------\n",
    "            #  Train Generator\n",
    "            # ---------------------\n",
    "\n",
    "            noise = np.random.normal(0, 1, (batch_size, 100))\n",
    "\n",
    "            # the generator wants the discriminator to label the generated samples\n",
    "            # as valid (ones)\n",
    "            valid_y = np.array([1] * 32)\n",
    "\n",
    "            # train the generator\n",
    "            g_loss = combined.train_on_batch(noise, np.ones((batch_size, 1)))\n",
    "\n",
    "            \n",
    "            # Plot the progress\n",
    "            print (\"%d [D loss: %f, acc.: %.2f%%] [G loss: %f]\" % (epoch, d_loss[0], 100*d_loss[1], g_loss))\n",
    "            D_L_REAL.append(d_loss_real)\n",
    "            D_L_FAKE.append(d_loss_fake)\n",
    "            D_L.append(d_loss)\n",
    "            D_ACC.append(d_loss[1])\n",
    "            G_L.append(g_loss)\n",
    "\n",
    "            # if at save interval => save generated image samples\n",
    "            if epoch % save_interval == 0:\n",
    "                save_imgs(dataset_title, epoch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "def save_imgs(dataset_title, epoch):\n",
    "        r, c = 5, 5\n",
    "        noise = np.random.normal(0, 1, (r * c, 100))\n",
    "        gen_imgs = generator.predict(noise)\n",
    "\n",
    "        # rescale images 0 - 1\n",
    "        gen_imgs = 0.5 * gen_imgs + 0.5\n",
    "\n",
    "        fig, axs = plt.subplots(r, c)\n",
    "        cnt = 0\n",
    "        \n",
    "        # iterate in order to create a subplot\n",
    "        for i in range(r):\n",
    "            for j in range(c):\n",
    "                if dataset_title == 'mnist':\n",
    "                    axs[i,j].imshow(gen_imgs[cnt, :,:,0], cmap='gray')\n",
    "                    axs[i,j].axis('off')\n",
    "                    cnt += 1\n",
    "                elif dataset_title == 'cifar10':\n",
    "                    axs[i,j].imshow(gen_imgs[cnt, :,:,:])\n",
    "                    axs[i,j].axis('off')\n",
    "                    cnt += 1\n",
    "                else:\n",
    "                    print('Please indicate the image options.')\n",
    "        \n",
    "        if not os.path.exists('images_{0}'.format(dataset_title)):\n",
    "            os.makedirs('images_{0}'.format(dataset_title))\n",
    "        \n",
    "        fig.savefig(\"images_{0}/{1}.png\".format(dataset_title, epoch))\n",
    "        plt.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "history = train('cifar10', epochs=30000, batch_size=32, save_interval=50)\n",
    "#generator.save('mnist_model.h5')\n",
    "generator.save('cifar10_model.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x86aa9cc278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(D_L)\n",
    "plt.title('Discriminator results (MNIST)')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Discriminator Loss (blue), Discriminator Accuracy (orange)')\n",
    "plt.legend(['Discriminator Loss', 'Discriminator Accuracy'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x86aaa0c048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(G_L)\n",
    "plt.title('Generator results (MNIST)')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Generator Loss (blue)')\n",
    "plt.legend('Generator Loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x654c8263c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(D_L)\n",
    "plt.title('Discriminator results (CIFAR10)')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Discriminator Loss (blue), Discriminator Accuracy (orange)')\n",
    "plt.legend(['Discriminator Loss', 'Discriminator Accuracy'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x654e895a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(G_L)\n",
    "plt.title('Generator results (CIFAR10)')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Generator Loss (blue)')\n",
    "plt.legend('Generator Loss')\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": "3.5.0"
  },
  "widgets": {
   "state": {},
   "version": "1.1.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}