examples/autoencoders/sparse.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "ce734de3",
   "metadata": {},
   "source": [
    "### Sparsity Constraint on the Encoded Representations\n",
    "\n",
    "Differences compared to the [previous example](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/intro.ipynb):  \n",
    "- *L1* regularizer has been included in the encoder dense layer\n",
    "- Number of epochs have been increased from 50 to 100 (it is possible without overtraining thanks to regularization)\n",
    "- *encoded_imgs.mean()* was $8.9$ in [intro](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/intro.ipynb). It is $1.3$ for this example thanks to regularization\n",
    "- *encoded_imgs.std()* was $4.6$ in [intro](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/intro.ipynb). It is $0.8$ for this example\n",
    "\n",
    "Previous example: [/examples/autoencoders/intro.ipynb](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/intro.ipynb)  \n",
    "Modified from: [blog.keras.io/building-autoencoders-in-keras](https://blog.keras.io/building-autoencoders-in-keras.html)  \n",
    "Next example: [/examples/autoencoders/convnet.ipynb](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/convnet.ipynb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "74e32580",
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras\n",
    "from keras import layers\n",
    "from keras.datasets import mnist\n",
    "from keras import regularizers\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e48d3a91",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Metal device set to: Apple M2\n",
      "\n",
      "systemMemory: 8.00 GB\n",
      "maxCacheSize: 2.67 GB\n",
      "\n",
      "autoencoder:\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_1 (InputLayer)        [(None, 784)]             0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 32)                25120     \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 784)               25872     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 50,992\n",
      "Trainable params: 50,992\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "\n",
      "encoder:\n",
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_1 (InputLayer)        [(None, 784)]             0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 32)                25120     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 25,120\n",
      "Trainable params: 25,120\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "\n",
      "decoder:\n",
      "Model: \"model_2\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_2 (InputLayer)        [(None, 32)]              0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 784)               25872     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 25,872\n",
      "Trainable params: 25,872\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-12-30 13:45:52.112578: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:306] Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.\n",
      "2022-12-30 13:45:52.112684: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:272] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)\n"
     ]
    }
   ],
   "source": [
    "# This is the size of our encoded representations\n",
    "encoding_dim = 32 # 32 floats -> compression of factor 24.5, assuming the input is 784 floats\n",
    "\n",
    "# This is our input image\n",
    "input_img = keras.Input(shape=(784,))\n",
    "# \"encoded\" is the encoded representation of the input\n",
    "encoded = layers.Dense(encoding_dim, activation='relu', activity_regularizer=regularizers.l1(10e-5))(input_img)\n",
    "# \"decoded\" is the lossy reconstruction of the input\n",
    "decoded = layers.Dense(784, activation='sigmoid')(encoded)\n",
    "\n",
    "# This model maps an input to its reconstruction\n",
    "autoencoder = keras.Model(input_img, decoded)\n",
    "\n",
    "# This model maps an input to its encoded representation\n",
    "encoder = keras.Model(input_img, encoded)\n",
    "\n",
    "# This is our encoded (32-dimensional) input\n",
    "encoded_input = keras.Input(shape=(encoding_dim,))\n",
    "# Retrieve the last layer of the autoencoder model\n",
    "decoder_layer = autoencoder.layers[-1]\n",
    "# Create the decoder model\n",
    "decoder = keras.Model(encoded_input, decoder_layer(encoded_input))\n",
    "\n",
    "autoencoder.compile(optimizer='adam', loss='binary_crossentropy')\n",
    "\n",
    "print('autoencoder:'); autoencoder.summary(); print()\n",
    "print('encoder:'); encoder.summary(); print()\n",
    "print('decoder:'); decoder.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "d1d5efa0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train.shape = (60000, 784)\n",
      "x_test.shape = (10000, 784)\n"
     ]
    }
   ],
   "source": [
    "(x_train, _), (x_test, _) = mnist.load_data() # Labels ignored\n",
    "x_train = x_train.astype('float32') / 255.\n",
    "x_test = x_test.astype('float32') / 255.\n",
    "x_train = x_train.reshape((len(x_train), np.prod(x_train.shape[1:])))\n",
    "x_test = x_test.reshape((len(x_test), np.prod(x_test.shape[1:])))\n",
    "print(f'{x_train.shape = }')\n",
    "print(f'{x_test.shape = }')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "0b95fd83",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-12-30 13:45:52.642517: W tensorflow/core/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz\n",
      "2022-12-30 13:45:52.814465: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "235/235 [==============================] - 3s 6ms/step - loss: 0.2867 - val_loss: 0.2019\n",
      "Epoch 2/100\n",
      "  1/235 [..............................] - ETA: 1s - loss: 0.2008"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-12-30 13:45:55.020297: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1809 - val_loss: 0.1632\n",
      "Epoch 3/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1544 - val_loss: 0.1442\n",
      "Epoch 4/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1394 - val_loss: 0.1324\n",
      "Epoch 5/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1295 - val_loss: 0.1241\n",
      "Epoch 6/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1224 - val_loss: 0.1182\n",
      "Epoch 7/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1174 - val_loss: 0.1141\n",
      "Epoch 8/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1139 - val_loss: 0.1112\n",
      "Epoch 9/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1115 - val_loss: 0.1091\n",
      "Epoch 10/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1096 - val_loss: 0.1076\n",
      "Epoch 11/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1083 - val_loss: 0.1064\n",
      "Epoch 12/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1073 - val_loss: 0.1056\n",
      "Epoch 13/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1065 - val_loss: 0.1048\n",
      "Epoch 14/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1058 - val_loss: 0.1043\n",
      "Epoch 15/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1053 - val_loss: 0.1037\n",
      "Epoch 16/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1048 - val_loss: 0.1033\n",
      "Epoch 17/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1043 - val_loss: 0.1029\n",
      "Epoch 18/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1039 - val_loss: 0.1025\n",
      "Epoch 19/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1036 - val_loss: 0.1022\n",
      "Epoch 20/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1033 - val_loss: 0.1019\n",
      "Epoch 21/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1030 - val_loss: 0.1016\n",
      "Epoch 22/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1027 - val_loss: 0.1014\n",
      "Epoch 23/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1024 - val_loss: 0.1011\n",
      "Epoch 24/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1022 - val_loss: 0.1009\n",
      "Epoch 25/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1020 - val_loss: 0.1007\n",
      "Epoch 26/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1018 - val_loss: 0.1005\n",
      "Epoch 27/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1016 - val_loss: 0.1003\n",
      "Epoch 28/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1014 - val_loss: 0.1002\n",
      "Epoch 29/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1012 - val_loss: 0.1000\n",
      "Epoch 30/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1010 - val_loss: 0.0998\n",
      "Epoch 31/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1009 - val_loss: 0.0997\n",
      "Epoch 32/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1007 - val_loss: 0.0996\n",
      "Epoch 33/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1006 - val_loss: 0.0994\n",
      "Epoch 34/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1005 - val_loss: 0.0993\n",
      "Epoch 35/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1003 - val_loss: 0.0992\n",
      "Epoch 36/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1002 - val_loss: 0.0990\n",
      "Epoch 37/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1001 - val_loss: 0.0990\n",
      "Epoch 38/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.1000 - val_loss: 0.0989\n",
      "Epoch 39/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0999 - val_loss: 0.0988\n",
      "Epoch 40/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0998 - val_loss: 0.0986\n",
      "Epoch 41/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0997 - val_loss: 0.0985\n",
      "Epoch 42/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0996 - val_loss: 0.0985\n",
      "Epoch 43/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0995 - val_loss: 0.0984\n",
      "Epoch 44/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0994 - val_loss: 0.0983\n",
      "Epoch 45/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0993 - val_loss: 0.0982\n",
      "Epoch 46/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0992 - val_loss: 0.0981\n",
      "Epoch 47/100\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.0991 - val_loss: 0.0980\n",
      "Epoch 48/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0990 - val_loss: 0.0980\n",
      "Epoch 49/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0990 - val_loss: 0.0979\n",
      "Epoch 50/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0989 - val_loss: 0.0978\n",
      "Epoch 51/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0988 - val_loss: 0.0977\n",
      "Epoch 52/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0988 - val_loss: 0.0977\n",
      "Epoch 53/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0987 - val_loss: 0.0977\n",
      "Epoch 54/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0986 - val_loss: 0.0976\n",
      "Epoch 55/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0986 - val_loss: 0.0975\n",
      "Epoch 56/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0985 - val_loss: 0.0975\n",
      "Epoch 57/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0985 - val_loss: 0.0974\n",
      "Epoch 58/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0984 - val_loss: 0.0973\n",
      "Epoch 59/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0983 - val_loss: 0.0973\n",
      "Epoch 60/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0983 - val_loss: 0.0972\n",
      "Epoch 61/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0982 - val_loss: 0.0972\n",
      "Epoch 62/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0982 - val_loss: 0.0972\n",
      "Epoch 63/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0981 - val_loss: 0.0971\n",
      "Epoch 64/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0981 - val_loss: 0.0971\n",
      "Epoch 65/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0980 - val_loss: 0.0970\n",
      "Epoch 66/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0980 - val_loss: 0.0969\n",
      "Epoch 67/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0979 - val_loss: 0.0969\n",
      "Epoch 68/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0979 - val_loss: 0.0969\n",
      "Epoch 69/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0978 - val_loss: 0.0968\n",
      "Epoch 70/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0978 - val_loss: 0.0967\n",
      "Epoch 71/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0977 - val_loss: 0.0967\n",
      "Epoch 72/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0977 - val_loss: 0.0967\n",
      "Epoch 73/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0977 - val_loss: 0.0967\n",
      "Epoch 74/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0976 - val_loss: 0.0967\n",
      "Epoch 75/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0976 - val_loss: 0.0966\n",
      "Epoch 76/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0975 - val_loss: 0.0966\n",
      "Epoch 77/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0975 - val_loss: 0.0966\n",
      "Epoch 78/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0975 - val_loss: 0.0965\n",
      "Epoch 79/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0974 - val_loss: 0.0965\n",
      "Epoch 80/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0974 - val_loss: 0.0963\n",
      "Epoch 81/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0974 - val_loss: 0.0963\n",
      "Epoch 82/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0973 - val_loss: 0.0964\n",
      "Epoch 83/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0973 - val_loss: 0.0963\n",
      "Epoch 84/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0972 - val_loss: 0.0963\n",
      "Epoch 85/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0972 - val_loss: 0.0962\n",
      "Epoch 86/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0972 - val_loss: 0.0963\n",
      "Epoch 87/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0971 - val_loss: 0.0962\n",
      "Epoch 88/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0971 - val_loss: 0.0961\n",
      "Epoch 89/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0971 - val_loss: 0.0962\n",
      "Epoch 90/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0971 - val_loss: 0.0960\n",
      "Epoch 91/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0970 - val_loss: 0.0960\n",
      "Epoch 92/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0970 - val_loss: 0.0960\n",
      "Epoch 93/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0970 - val_loss: 0.0960\n",
      "Epoch 94/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0969 - val_loss: 0.0959\n",
      "Epoch 95/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0969 - val_loss: 0.0959\n",
      "Epoch 96/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0969 - val_loss: 0.0959\n",
      "Epoch 97/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0968 - val_loss: 0.0959\n",
      "Epoch 98/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0968 - val_loss: 0.0958\n",
      "Epoch 99/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0968 - val_loss: 0.0958\n",
      "Epoch 100/100\n",
      "235/235 [==============================] - 1s 5ms/step - loss: 0.0968 - val_loss: 0.0958\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1679cab90>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "autoencoder.fit(x_train, x_train, epochs=100, batch_size=256, shuffle=True, validation_data=(x_test, x_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "0ad16e1a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-12-30 13:47:49.825106: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.\n",
      "2022-12-30 13:47:50.157988: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "encoded_imgs.mean() = 1.2759832\n",
      "encoded_imgs.std() = 0.8427344\n"
     ]
    }
   ],
   "source": [
    "# Encode and decode some digits\n",
    "# Note that we take them from the *test* set\n",
    "encoded_imgs = encoder.predict(x_test, verbose=0)\n",
    "decoded_imgs = decoder.predict(encoded_imgs, verbose=0)\n",
    "print(f'{encoded_imgs.mean() = }')\n",
    "print(f'{encoded_imgs.std() = }')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c2ab5fea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x200 with 20 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "num_of_digits_to_display = 10\n",
    "plt.style.use('dark_background')\n",
    "fig = plt.figure(figsize=(10, 2))\n",
    "fig.subplots_adjust(hspace=0)\n",
    "for digit_idx in range(num_of_digits_to_display):\n",
    "    # Display original\n",
    "    ax = plt.subplot(2, num_of_digits_to_display, digit_idx + 1)\n",
    "    plt.imshow(x_test[digit_idx].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "\n",
    "    # Display reconstruction\n",
    "    ax = plt.subplot(2, num_of_digits_to_display, digit_idx + 1 + num_of_digits_to_display)\n",
    "    plt.imshow(decoded_imgs[digit_idx].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "c484cb68",
   "metadata": {},
   "source": [
    "Previous example: [/examples/autoencoders/intro.ipynb](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/intro.ipynb)  \n",
    "Modified from: [blog.keras.io/building-autoencoders-in-keras](https://blog.keras.io/building-autoencoders-in-keras.html)  \n",
    "Next example: [/examples/autoencoders/convnet.ipynb](https://github.com/serhatsoyer/py4ML/blob/main/examples/autoencoders/convnet.ipynb)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.6 ('study')",
   "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": "3.10.8 (main, Nov 24 2022, 08:08:27) [Clang 14.0.6 ]"
  },
  "vscode": {
   "interpreter": {
    "hash": "29d06b973d1ddb34db3279b24f9b5152402e688db937648b736e855ec4de60c0"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}