Auto_encoder_demo.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Auto_encoder_demo.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyM9+XE+tetvCxj7ZbCAFAvh",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/novoforce/Exploring-Tensorflow/blob/main/Auto_encoder_demo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HKkS6hrjoely",
        "outputId": "c5facf39-f00a-4012-f114-892347cd96d1"
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from keras.layers import Dense,Input\n",
        "from keras.models import Model\n",
        "\n",
        "print(tf.__version__)"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2.6.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MkJGFy0dqo9c"
      },
      "source": [
        "# Vanilla Auto-encoder"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IRvh4p9To0hW",
        "outputId": "74a9b3e4-43ff-4db0-b2f5-cd92d522918e"
      },
      "source": [
        "# 784-> 32-> 784\n",
        "encoding_dim= 32  # compress the image to the factor of 784/32 == 24.5\n",
        "input_img= Input(shape=(784,))\n",
        "encoded= Dense(encoding_dim,activation=\"relu\")(input_img) #encoded representation\n",
        "decoded= Dense(784,activation=\"sigmoid\")(encoded) #reconstruction from the encoded form\n",
        "\n",
        "#The reason behind usage of sigmoid is so as to get the image in the range of 0-1(normalized)\n",
        "auto_encoder= Model(input_img,decoded)\n",
        "\n",
        "auto_encoder.summary()"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "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",
            "Total params: 50,992\n",
            "Trainable params: 50,992\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "h8Y3nj_AqWfh",
        "outputId": "77549893-bff4-4189-8b8f-cd919763c275"
      },
      "source": [
        "# Build encoder model\n",
        "encoder= Model(input_img,encoded)\n",
        "encoder.summary()"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "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",
            "Total params: 25,120\n",
            "Trainable params: 25,120\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wT-AoYzUr71x",
        "outputId": "d14d5cc1-0ce9-48a7-f861-b3727f1d1d85"
      },
      "source": [
        "# Build decoder\n",
        "encoded_input= Input(shape=(encoding_dim,)) # flexibility of giving our own input\n",
        "decoder_layer= auto_encoder.layers[-1] (encoded_input) #retreiving the last layer of the autoencoder(to keep the trained weights)\n",
        "decoder= Model(encoded_input,decoder_layer)\n",
        "\n",
        "decoder.summary()"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "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",
            "Total params: 25,872\n",
            "Trainable params: 25,872\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QWsoh7F8y_j9"
      },
      "source": [
        "# Pre-process Mnist data\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "S7v9cunLtJa3",
        "outputId": "375179ab-74f7-41a5-a20d-1c2d4a2a794e"
      },
      "source": [
        "(x_train,_), (x_test, _) = tf.keras.datasets.mnist.load_data()"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 0s 0us/step\n",
            "11501568/11490434 [==============================] - 0s 0us/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WF4VK2ZdzLJZ",
        "outputId": "02e6d592-273d-4afe-e120-e2932830ee32"
      },
      "source": [
        "import numpy as np\n",
        "\n",
        "x_train= x_train.astype('float32') / 255\n",
        "x_test= x_test.astype('float32') / 255\n",
        "\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",
        "\n",
        "print(x_train.shape, x_test.shape)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(60000, 784) (10000, 784)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "g7kxCymq0MJr",
        "outputId": "4e66aaaf-6226-4896-f3fe-a42c1db7d098"
      },
      "source": [
        "auto_encoder.compile(optimizer='adam',loss='binary_crossentropy')\n",
        "auto_encoder.fit(x_train,x_train,epochs=25,batch_size=256,shuffle=True,validation_data=(x_test,x_test))"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/25\n",
            "235/235 [==============================] - 4s 5ms/step - loss: 0.2738 - val_loss: 0.1892\n",
            "Epoch 2/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.1714 - val_loss: 0.1544\n",
            "Epoch 3/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.1446 - val_loss: 0.1335\n",
            "Epoch 4/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.1286 - val_loss: 0.1214\n",
            "Epoch 5/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.1186 - val_loss: 0.1132\n",
            "Epoch 6/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.1117 - val_loss: 0.1076\n",
            "Epoch 7/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.1069 - val_loss: 0.1035\n",
            "Epoch 8/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.1032 - val_loss: 0.1002\n",
            "Epoch 9/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.1002 - val_loss: 0.0976\n",
            "Epoch 10/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0980 - val_loss: 0.0959\n",
            "Epoch 11/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.0965 - val_loss: 0.0946\n",
            "Epoch 12/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0956 - val_loss: 0.0939\n",
            "Epoch 13/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0949 - val_loss: 0.0933\n",
            "Epoch 14/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0945 - val_loss: 0.0931\n",
            "Epoch 15/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.0942 - val_loss: 0.0928\n",
            "Epoch 16/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0940 - val_loss: 0.0926\n",
            "Epoch 17/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0938 - val_loss: 0.0925\n",
            "Epoch 18/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0937 - val_loss: 0.0924\n",
            "Epoch 19/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.0936 - val_loss: 0.0923\n",
            "Epoch 20/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.0935 - val_loss: 0.0922\n",
            "Epoch 21/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0934 - val_loss: 0.0921\n",
            "Epoch 22/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0933 - val_loss: 0.0921\n",
            "Epoch 23/25\n",
            "235/235 [==============================] - 1s 5ms/step - loss: 0.0933 - val_loss: 0.0921\n",
            "Epoch 24/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.0932 - val_loss: 0.0920\n",
            "Epoch 25/25\n",
            "235/235 [==============================] - 1s 4ms/step - loss: 0.0932 - val_loss: 0.0920\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f0f830ea050>"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V9BIoQv_1C3U"
      },
      "source": [
        "encoded_img= encoder.predict(x_test)\n",
        "decoded_img= decoder.predict(encoded_img)"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "69gKMwpx8GJC"
      },
      "source": [
        "# X_test "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "id": "GVj4P0uW7GEw",
        "outputId": "90f12419-17c3-4d20-9c8c-befc51485436"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.figure(figsize=(12, 12))\n",
        "for i in range(0,9):\n",
        "    plt.subplot(5, 3, i+1)\n",
        "    plt.imshow(x_test[i].reshape(28,28), cmap=\"gray\") #'image_list' is the list of images\n",
        "    # plt.xlabel(CLASSES[labels[i]]) # 'labels' is the list of labels\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x864 with 9 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "paxemKpR8DYb"
      },
      "source": [
        "# Decoded output"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "id": "EkCxHzOY7chz",
        "outputId": "8ba43e6c-2c32-4fab-e792-82d640df8717"
      },
      "source": [
        "\n",
        "plt.figure(figsize=(12, 12))\n",
        "for i in range(0,9):\n",
        "    plt.subplot(5, 3, i+1)\n",
        "    plt.imshow(decoded_img[i].reshape(28,28), cmap=\"gray\") #'image_list' is the list of images\n",
        "    # plt.xlabel(CLASSES[labels[i]]) # 'labels' is the list of labels\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x864 with 9 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y4Ua2uc08zbF"
      },
      "source": [
        "# Save the models"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pSGj6U5H7qqp",
        "outputId": "a250f8b8-e4c8-4ed7-e5ef-dd5e2eefee71"
      },
      "source": [
        "encoder.save(\"/content\")\n",
        "decoder.save(\"/content\")"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            "INFO:tensorflow:Assets written to: /content/assets\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            "INFO:tensorflow:Assets written to: /content/assets\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UDRGyAfk83FO"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}