convolutional beta-VAE on F_MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# VAE model building"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import module"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import Lambda,Input,Dense,Flatten,Conv2D,Conv2DTranspose\n",
    "from keras.layers import Activation,BatchNormalization,Reshape\n",
    "from keras.models import Model\n",
    "from keras.datasets import fashion_mnist,mnist\n",
    "from keras.losses import mse, binary_crossentropy\n",
    "from keras.callbacks import Callback,ModelCheckpoint,ReduceLROnPlateau\n",
    "from keras import backend as K\n",
    "from keras.optimizers import Adam\n",
    "from keras import initializers\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import cv2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sampling(arg):\n",
    "    arg = [z_mean,z_log_var]\n",
    "    dim = K.int_shape(z_mean)[1]\n",
    "    epsilon = K.random_normal(shape=(K.shape(z_mean)[0], dim),mean=0.0, stddev=1.0) # reparameterization trick\n",
    "    return z_mean + K.exp(0.5 * z_log_var) * epsilon"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "(X_train, y_train), (X_test, y_test) = fashion_mnist.load_data()\n",
    "fmnist_label = {0:'T-shirt/top',1:'Trouser',2:'Pullover',3:'Dress',4:'Coat',5:'Sandal',6:'Shirt',7:'Sneaker',8:'Bag',9:'Ankle boot'}\n",
    "y_train = list(map(lambda x:fmnist_label[x], y_train))\n",
    "y_test = list(map(lambda x:fmnist_label[x], y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x2160 with 50 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_samples = 5\n",
    "n_cls = len(np.unique(y_train))\n",
    "fig, rows = plt.subplots(n_cls,n_samples,figsize=(4*n_samples,3*n_cls))\n",
    "\n",
    "for row,cls in zip(rows,range(n_cls)):\n",
    "    row[int(np.floor(n_samples/2))].set_title(fmnist_label[cls],fontsize=25)\n",
    "    for col_ax,img_idx in zip(row,[i for i,v in enumerate(y_train) if v==fmnist_label[cls]][:n_samples]):\n",
    "        img = X_train[img_idx]\n",
    "        col_ax.imshow(img,cmap='gray')\n",
    "        col_ax.axis('off')\n",
    "plt.subplots_adjust(left=0.2, wspace=0.02)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape: (60000, 28, 28, 1)\n",
      "X_test shape: (10000, 28, 28, 1)\n"
     ]
    }
   ],
   "source": [
    "X_train = np.expand_dims(X_train,-1)\n",
    "X_test =  np.expand_dims(X_test,-1)\n",
    "X_train = X_train.astype('float32')/127.5-1 # for tanh\n",
    "X_test = X_test.astype('float32')/127.5-1\n",
    "print('X_train shape: {}'.format(X_train.shape))\n",
    "print('X_test shape: {}'.format(X_test.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "channel = 1\n",
    "img_rows = X_train.shape[1]\n",
    "img_cols = X_train.shape[2]\n",
    "input_shape = (img_rows,img_cols,channel)\n",
    "batch_size = 128\n",
    "latent_dim = 30\n",
    "epochs = 100\n",
    "learning_rate = 0.001\n",
    "learning_rate_reduction = ReduceLROnPlateau(monitor='val_loss',patience=10,verbose=1,factor=0.5,min_lr=0.00001)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Encoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            (None, 28, 28, 1)    0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1 (Conv2D)               (None, 28, 28, 16)   160         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1 (BatchNor (None, 28, 28, 16)   64          conv2d_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1 (Activation)       (None, 28, 28, 16)   0           batch_normalization_1[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_2 (Conv2D)               (None, 14, 14, 32)   4640        activation_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_2 (BatchNor (None, 14, 14, 32)   128         conv2d_2[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "activation_2 (Activation)       (None, 14, 14, 32)   0           batch_normalization_2[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_3 (Conv2D)               (None, 7, 7, 64)     18496       activation_2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_3 (BatchNor (None, 7, 7, 64)     256         conv2d_3[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "activation_3 (Activation)       (None, 7, 7, 64)     0           batch_normalization_3[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)             (None, 3136)         0           activation_3[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "flatten_2 (Flatten)             (None, 3136)         0           activation_3[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "z_mean (Dense)                  (None, 30)           94110       flatten_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "z_log_var (Dense)               (None, 30)           94110       flatten_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_4 (BatchNor (None, 30)           120         z_mean[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_5 (BatchNor (None, 30)           120         z_log_var[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_1 (Lambda)               (None, 30)           0           batch_normalization_4[0][0]      \n",
      "                                                                 batch_normalization_5[0][0]      \n",
      "==================================================================================================\n",
      "Total params: 212,204\n",
      "Trainable params: 211,860\n",
      "Non-trainable params: 344\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "w_init = initializers.random_normal(stddev=0.02)\n",
    "gamma_init = initializers.random_normal(mean=1.0,stddev=0.02)\n",
    "\n",
    "inputs = Input(shape=input_shape)\n",
    "\n",
    "x = Conv2D(16,3,strides=1,padding='same',kernel_initializer=w_init)(inputs)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "x = Activation('relu')(x)\n",
    "x = Conv2D(32,3,strides=2,padding='same',kernel_initializer=w_init)(x)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "x = Activation('relu')(x)\n",
    "x = Conv2D(64,3,strides=2,padding='same',kernel_initializer=w_init)(x)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "x = Activation('relu')(x)\n",
    "before_flatten_shape = K.int_shape(x)\n",
    "\n",
    "x_z = Flatten()(x)\n",
    "z_mean = Dense(latent_dim,kernel_initializer=w_init,name='z_mean')(x_z)\n",
    "z_mean = BatchNormalization(gamma_initializer=gamma_init)(z_mean)\n",
    "\n",
    "x_var = Flatten()(x)\n",
    "z_log_var = Dense(latent_dim,kernel_initializer=w_init,name='z_log_var')(x_var)\n",
    "z_log_var = BatchNormalization(gamma_initializer=gamma_init)(z_log_var)\n",
    "\n",
    "# z_mean = Dense(latent_dim, name='z_mean')(x)\n",
    "# z_log_var = Dense(latent_dim, name='z_log_var')(x)\n",
    "\n",
    "z = Lambda(sampling, output_shape=(latent_dim,))([z_mean, z_log_var])\n",
    "\n",
    "encoder = Model(inputs, [z_mean,z_log_var,z])\n",
    "encoder.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Decoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         (None, 30)                0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 3136)              97216     \n",
      "_________________________________________________________________\n",
      "reshape_1 (Reshape)          (None, 7, 7, 64)          0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 7, 7, 64)          256       \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_1 (Conv2DTr (None, 14, 14, 64)        36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_7 (Batch (None, 14, 14, 64)        256       \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 14, 14, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_2 (Conv2DTr (None, 28, 28, 32)        18464     \n",
      "_________________________________________________________________\n",
      "batch_normalization_8 (Batch (None, 28, 28, 32)        128       \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 28, 28, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_3 (Conv2DTr (None, 28, 28, 16)        4624      \n",
      "_________________________________________________________________\n",
      "batch_normalization_9 (Batch (None, 28, 28, 16)        64        \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 28, 28, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_4 (Conv2DTr (None, 28, 28, 1)         145       \n",
      "=================================================================\n",
      "Total params: 158,081\n",
      "Trainable params: 157,729\n",
      "Non-trainable params: 352\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "latent_inputs = Input(shape=(latent_dim,))\n",
    "x = Dense(before_flatten_shape[1]*before_flatten_shape[2]*before_flatten_shape[3],activation='relu',kernel_initializer=w_init)(latent_inputs)\n",
    "x = Reshape((before_flatten_shape[1],before_flatten_shape[2],before_flatten_shape[3]))(x)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "\n",
    "x = Conv2DTranspose(64,3,strides=2,padding='same',kernel_initializer=w_init)(x)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "x = Activation('relu')(x)\n",
    "\n",
    "x = Conv2DTranspose(32,3,strides=2,padding='same',kernel_initializer=w_init)(x)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "x = Activation('relu')(x)\n",
    "\n",
    "x = Conv2DTranspose(16,3,strides=1,padding='same',kernel_initializer=w_init)(x)\n",
    "x = BatchNormalization(gamma_initializer=gamma_init)(x)\n",
    "x = Activation('relu')(x)\n",
    "\n",
    "outputs = Conv2DTranspose(channel,3,activation='tanh',padding='same',kernel_initializer=w_init)(x)\n",
    "\n",
    "# instantiate decoder model\n",
    "decoder = Model(latent_inputs, outputs)\n",
    "decoder.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## VAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "outputs = decoder(encoder(inputs)[2])\n",
    "vae = Model(inputs, outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "beta = 1 # 1 --> regular VAE\n",
    "reconstruction_loss = mse(K.flatten(inputs), K.flatten(outputs))\n",
    "# reconstruction_loss = binary_crossentropy(K.flatten(inputs),K.flatten(outputs))\n",
    "reconstruction_loss *= img_rows*img_cols\n",
    "\n",
    "kl_loss = -0.5*beta*K.sum(1+z_log_var-K.square(z_mean)-K.exp(z_log_var),axis=-1)\n",
    "vae_loss = K.mean(reconstruction_loss+kl_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.makedirs('best_weight_vae_v1',exist_ok=True)\n",
    "best_model_weight_path = os.path.join('best_weight_vae_v1','vae_v1'+'-best-wiehgts'+'-{epoch:03d}-{loss:.3f}-{val_loss:.3f}.h5')\n",
    "save_best_model = ModelCheckpoint(best_model_weight_path,monitor='val_loss',verbose=0,save_weights_only=True,save_best_only=True,mode='min')\n",
    "\n",
    "class NEpochPrint(Callback):\n",
    "    def __init__(self, display_step):\n",
    "        self.epoch = 0\n",
    "        self.display_step = display_step\n",
    "    def on_epoch_end(self,epoch,logs={}):\n",
    "        self.epoch += 1        \n",
    "        if self.epoch == 1 or self.epoch % self.display_step == 0:\n",
    "            print('Epoch: {}/{} ..... {}: {:.4f} - {}: {:.4f}'.format(self.epoch,\n",
    "                                                                      self.params['epochs'],\n",
    "                                                                      self.params['metrics'][0], \n",
    "                                                                      logs.get(self.params['metrics'][0]),\n",
    "                                                                      self.params['metrics'][1],               \n",
    "                                                                      logs.get(self.params['metrics'][1])))\n",
    "NEpochPrinter = NEpochPrint(display_step=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_or_load_weights(flag):\n",
    "    if flag == 'train':\n",
    "        hist = vae.fit(X_train,\n",
    "                       epochs=epochs,\n",
    "                       batch_size=batch_size,\n",
    "                       validation_data=(X_test, None),\n",
    "                       verbose=0,\n",
    "                       callbacks=[save_best_model,learning_rate_reduction,NEpochPrinter])\n",
    "        vae.save_weights('vae_fashion_mnist.h5')\n",
    "        return hist\n",
    "    if flag == 'load':\n",
    "        vae.load_weights('vae_mnist.h5') # load all the weights for encoder and decoder when loading for vae"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 28, 28, 1)         0         \n",
      "_________________________________________________________________\n",
      "model_1 (Model)              [(None, 30), (None, 30),  212204    \n",
      "_________________________________________________________________\n",
      "model_2 (Model)              (None, 28, 28, 1)         158081    \n",
      "=================================================================\n",
      "Total params: 370,285\n",
      "Trainable params: 369,589\n",
      "Non-trainable params: 696\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "vae.add_loss(vae_loss)\n",
    "vae.compile(optimizer=Adam(lr=learning_rate))\n",
    "vae.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1/100 ..... loss: 110.5745 - val_loss: 95.9095\n",
      "Epoch: 5/100 ..... loss: 69.2447 - val_loss: 70.0717\n",
      "Epoch: 10/100 ..... loss: 65.2837 - val_loss: 65.1178\n",
      "Epoch: 15/100 ..... loss: 63.5700 - val_loss: 63.3591\n",
      "Epoch: 20/100 ..... loss: 62.4216 - val_loss: 61.3510\n",
      "Epoch: 25/100 ..... loss: 61.5829 - val_loss: 60.9745\n",
      "Epoch: 30/100 ..... loss: 60.9822 - val_loss: 60.4954\n",
      "Epoch: 35/100 ..... loss: 60.4390 - val_loss: 60.0123\n",
      "Epoch: 40/100 ..... loss: 60.0419 - val_loss: 59.8472\n",
      "Epoch: 45/100 ..... loss: 59.8021 - val_loss: 59.3311\n",
      "Epoch: 50/100 ..... loss: 59.4235 - val_loss: 58.8687\n",
      "Epoch: 55/100 ..... loss: 59.2181 - val_loss: 58.6516\n",
      "Epoch: 60/100 ..... loss: 58.9410 - val_loss: 58.7766\n",
      "Epoch: 65/100 ..... loss: 58.8374 - val_loss: 58.9098\n",
      "\n",
      "Epoch 00067: ReduceLROnPlateau reducing learning rate to 0.0005000000237487257.\n",
      "Epoch: 70/100 ..... loss: 58.0774 - val_loss: 57.8873\n",
      "Epoch: 75/100 ..... loss: 57.9590 - val_loss: 57.6100\n",
      "Epoch: 80/100 ..... loss: 57.8463 - val_loss: 57.5384\n",
      "Epoch: 85/100 ..... loss: 57.7980 - val_loss: 57.5793\n",
      "Epoch: 90/100 ..... loss: 57.7071 - val_loss: 57.4850\n",
      "Epoch: 95/100 ..... loss: 57.6773 - val_loss: 57.3639\n",
      "Epoch: 100/100 ..... loss: 57.5544 - val_loss: 57.4012\n"
     ]
    }
   ],
   "source": [
    "hist = train_or_load_weights(flag='train')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def training_plot(hist,model_name):\n",
    "    loss = [hist.history['loss'],hist.history['val_loss']]\n",
    "    plt.plot(loss[0],color='b',label='training')\n",
    "    plt.plot(loss[1],color='r',label='valid')\n",
    "    plt.title(model_name,fontsize=18)\n",
    "    plt.legend(loc='best',fontsize=13)\n",
    "    plt.xlabel('epoch',fontsize=15)\n",
    "    plt.ylabel('')\n",
    "    plt.tick_params(axis='both',labelsize=13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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"
    }
   ],
   "source": [
    "training_plot(hist,'VAE on f_MNIST')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_five_reconstruction(input_set,random_idx,encoder,decoder):\n",
    "    fig, axes = plt.subplots(2,5,figsize=(15,5))\n",
    "    for row,flag in zip(axes,['in_img','out_img']): \n",
    "        for col,idx in zip(row,random_idx):\n",
    "            original = input_set[idx]\n",
    "            if flag == 'in_img':\n",
    "                original_digit = original.reshape(img_rows,img_cols)\n",
    "                col.imshow(original_digit,cmap='bone')\n",
    "                col.set_title('original, idx={}'.format(idx))\n",
    "            if flag == 'out_img':\n",
    "                encoded_z = encoder.predict(np.expand_dims(original,0),batch_size=1)[2]\n",
    "                predict = decoder.predict(encoded_z)\n",
    "                decoded_digit = predict.reshape(img_rows,img_cols)\n",
    "                col.imshow(decoded_digit,cmap='bone')\n",
    "                col.set_title('reconstructed, idx={}'.format(idx))\n",
    "            col.axis('off')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Traing set reconstruction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.random.seed(100)\n",
    "random_idx = np.random.randint(len(X_train),size=5)\n",
    "plot_five_reconstruction(X_train,random_idx,encoder,decoder)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test set reconstruction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.random.seed(100)\n",
    "random_idx = np.random.randint(len(X_test),size=5)\n",
    "plot_five_reconstruction(X_test,random_idx,encoder,decoder)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Interpolation on the test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# X_test_encoded = encoder.predict(X_test, batch_size=1)[2]\n",
    "# latent_space_df = pd.DataFrame(X_test_encoded,columns=['vector1','vector2'])\n",
    "# latent_space_df['label'] = y_test\n",
    "# latent_space_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sns.set_context('paper', font_scale=1.5)\n",
    "# sns.set_style('white')\n",
    "# plt.figure()\n",
    "# sns.lmplot('vector1','vector2',data=latent_space_df,hue='label',fit_reg=False,scatter_kws={'alpha':0.6,\"s\": 40},\n",
    "#             height=8,palette=sns.color_palette(\"hls\",latent_space_df['label'].nunique()))\n",
    "# plt.title('Latent space by class')\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# n = 30\n",
    "# digit_size = 28\n",
    "# figure = np.zeros((digit_size * n, digit_size * n))\n",
    "\n",
    "# grid_x = np.linspace(-4, 4, n)\n",
    "# grid_y = np.linspace(-4, 4, n)\n",
    "\n",
    "# for i, yi in enumerate(grid_x):\n",
    "#     for j, xi in enumerate(grid_y):\n",
    "#         z_sample = np.array([[xi, yi]])\n",
    "#         x_decoded = decoder.predict(z_sample)\n",
    "#         digit = x_decoded[0].reshape(digit_size, digit_size)\n",
    "#         figure[i * digit_size: (i + 1) * digit_size,\n",
    "#                j * digit_size: (j + 1) * digit_size] = digit\n",
    "# plt.figure(figsize=(10, 10))\n",
    "# plt.imshow(figure,cmap='bone')\n",
    "# plt.axis('off')\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def interpolation(img1,img2,step,encoder,decoder,plot=True):\n",
    "    img1_encoded_z = encoder.predict(np.expand_dims(img1,0),batch_size=1)[2]\n",
    "#     print('1: {}'.format(img1_encoded_z))   \n",
    "    img2_encoded_z = encoder.predict(np.expand_dims(img2,0),batch_size=1)[2]\n",
    "#     print('2: {}'.format(img2_encoded_z))\n",
    "    step_size = (img2_encoded_z-img1_encoded_z)/step\n",
    "#     print('step: {}'.format(step_size))\n",
    "\n",
    "    step_list = [img1_encoded_z]\n",
    "    for _ in range(step):\n",
    "        img1_encoded_z = img1_encoded_z + step_size\n",
    "        step_list.append(img1_encoded_z)\n",
    "    if plot:\n",
    "        os.makedirs('interpolation_images',exist_ok=True)\n",
    "        fig,axes = plt.subplots(int(np.ceil((step+1)/5)),5, figsize=(5*4, int(np.ceil((step+1)/5))*4))\n",
    "        for i,(ax,vector) in enumerate(zip(axes.flatten(),step_list)):\n",
    "            predict = decoder.predict(vector).reshape(img_rows,img_cols)\n",
    "            ax.imshow(predict,cmap='bone')\n",
    "            ax.axis('off')\n",
    "            \n",
    "            fig_indi,ax_indi = plt.subplots(figsize=(5,5))\n",
    "            ax_indi.imshow(predict,cmap='bone')\n",
    "            ax_indi.axis('off')\n",
    "            plt.title('step {}/{}'.format(i,step),fontsize=24,fontweight='bold')\n",
    "            fig_indi.savefig('interpolation_images/step_{}.png'.format(i))\n",
    "            plt.close()\n",
    "            if i == 0:\n",
    "                ax.set_title('From Image 1\\n0%',fontsize=24,fontweight='bold')\n",
    "            elif i == step:\n",
    "                ax.set_title('To Image 2\\n100%',fontsize=24,fontweight='bold')\n",
    "            else:\n",
    "                ax.set_title('step {}/{}\\n{:.0f}%'.format(i,step,i*100/step),fontsize=24,fontweight='bold')\n",
    "        plt.subplots_adjust(hspace=0.3,left=0.1, wspace=0.02)\n",
    "        if (step+1)%5 != 0:\n",
    "            for i in axes.flatten()[(step+1)-len(axes.flatten()):]:\n",
    "                i.set_visible(False)\n",
    "        fig.savefig('interpolation_images/summary.png')\n",
    "    return step_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 25 496]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x864 with 15 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.random.seed(150)\n",
    "random_idx_2 = np.random.randint(len(X_test),size=2)\n",
    "print(random_idx_2)\n",
    "_ = interpolation(X_test[random_idx_2[0]],X_test[random_idx_2[1]],10,encoder,decoder,plot=True)"
   ]
  }
 ],
 "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
}