5_Demo_Keras_Autoenconder.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "5_Demo_Keras_Autoenconder.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "metadata": {
        "id": "PgHG5JaR_RdW",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Demo 5 (Bônus)  - Autoencoders usando Keras e MNIST"
      ]
    },
    {
      "metadata": {
        "id": "pqE2vueev9BB",
        "colab_type": "code",
        "outputId": "4a221d9d-8059-4cd3-9b68-1f9a981da531",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "import keras\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "\n",
        "from keras.datasets import mnist\n",
        "from keras.models import Model\n",
        "from keras.layers import Input, add\n",
        "from keras.layers.core import Layer, Dense, Dropout, Activation, Flatten, Reshape\n",
        "from keras import regularizers\n",
        "from keras.regularizers import l2\n",
        "from keras.layers.convolutional import Conv2D, MaxPooling2D, UpSampling2D, ZeroPadding2D\n",
        "from keras.utils import np_utils"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "p6dKwjk5xxq4",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "(X_train, _), (X_test, _) = mnist.load_data()\n",
        "\n",
        "X_train = X_train.reshape(X_train.shape[0], 28, 28, 1)\n",
        "X_test = X_test.reshape(X_test.shape[0], 28, 28, 1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "IZojdMy3x1EZ",
        "colab_type": "code",
        "outputId": "e4b35331-aadd-4299-e686-99d865dd6541",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "X_train = X_train.astype(\"float32\")/255.\n",
        "X_test = X_test.astype(\"float32\")/255.\n",
        "\n",
        "print('X_train shape:', X_train.shape)\n",
        "print(X_train.shape[0], 'train samples')\n",
        "print(X_test.shape[0], 'test samples')"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X_train shape: (60000, 28, 28, 1)\n",
            "60000 train samples\n",
            "10000 test samples\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "vJhyC3gmx4z0",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "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:])))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "9nrHiUdVx6vA",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "input_size = 784\n",
        "hidden_size = 64\n",
        "output_size = 784"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ukYRUKBZx-ea",
        "colab_type": "code",
        "outputId": "15f34405-f428-40bd-9f0a-410619a07f93",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "x = Input(shape=(input_size,))\n",
        "h = Dense(hidden_size, activation='relu')(x)\n",
        "r = Dense(output_size, activation='sigmoid')(h)\n",
        "\n",
        "autoencoder = Model(inputs=x, outputs=r)\n",
        "autoencoder.compile(optimizer='adam', loss='mse')"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "32dOss6oyA7S",
        "colab_type": "code",
        "outputId": "5939e906-cbb9-4c11-981d-f54c3d0708fb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 276
        }
      },
      "cell_type": "code",
      "source": [
        "from IPython.display import SVG\n",
        "from keras.utils.vis_utils import model_to_dot\n",
        "\n",
        "SVG(model_to_dot(autoencoder).create(prog='dot', format='svg'))"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<IPython.core.display.SVG object>"
            ],
            "image/svg+xml": "<svg height=\"191pt\" viewBox=\"0.00 0.00 141.00 191.00\" width=\"141pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n<g class=\"graph\" id=\"graph0\" transform=\"scale(1 1) rotate(0) translate(4 187)\">\n<title>G</title>\n<polygon fill=\"#ffffff\" points=\"-4,4 -4,-187 137,-187 137,4 -4,4\" stroke=\"transparent\"/>\n<!-- 140320526584184 -->\n<g class=\"node\" id=\"node1\">\n<title>140320526584184</title>\n<polygon fill=\"none\" points=\"0,-146.5 0,-182.5 133,-182.5 133,-146.5 0,-146.5\" stroke=\"#000000\"/>\n<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"66.5\" y=\"-160.8\">input_1: InputLayer</text>\n</g>\n<!-- 140320526584632 -->\n<g class=\"node\" id=\"node2\">\n<title>140320526584632</title>\n<polygon fill=\"none\" points=\"13,-73.5 13,-109.5 120,-109.5 120,-73.5 13,-73.5\" stroke=\"#000000\"/>\n<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"66.5\" y=\"-87.8\">dense_1: Dense</text>\n</g>\n<!-- 140320526584184&#45;&gt;140320526584632 -->\n<g class=\"edge\" id=\"edge1\">\n<title>140320526584184-&gt;140320526584632</title>\n<path d=\"M66.5,-146.4551C66.5,-138.3828 66.5,-128.6764 66.5,-119.6817\" fill=\"none\" stroke=\"#000000\"/>\n<polygon fill=\"#000000\" points=\"70.0001,-119.5903 66.5,-109.5904 63.0001,-119.5904 70.0001,-119.5903\" stroke=\"#000000\"/>\n</g>\n<!-- 140320526585248 -->\n<g class=\"node\" id=\"node3\">\n<title>140320526585248</title>\n<polygon fill=\"none\" points=\"13,-.5 13,-36.5 120,-36.5 120,-.5 13,-.5\" stroke=\"#000000\"/>\n<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"66.5\" y=\"-14.8\">dense_2: Dense</text>\n</g>\n<!-- 140320526584632&#45;&gt;140320526585248 -->\n<g class=\"edge\" id=\"edge2\">\n<title>140320526584632-&gt;140320526585248</title>\n<path d=\"M66.5,-73.4551C66.5,-65.3828 66.5,-55.6764 66.5,-46.6817\" fill=\"none\" stroke=\"#000000\"/>\n<polygon fill=\"#000000\" points=\"70.0001,-46.5903 66.5,-36.5904 63.0001,-46.5904 70.0001,-46.5903\" stroke=\"#000000\"/>\n</g>\n</g>\n</svg>"
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "metadata": {
        "id": "gasUH7IcyD8v",
        "colab_type": "code",
        "outputId": "3d52d990-54a0-4cd8-f8a6-2a96606ba27c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "cell_type": "code",
      "source": [
        "epochs = 5\n",
        "batch_size = 128\n",
        "\n",
        "history = autoencoder.fit(X_train, X_train, batch_size=batch_size, epochs=epochs, verbose=1, validation_data=(X_test, X_test))"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 60000 samples, validate on 10000 samples\n",
            "Epoch 1/5\n",
            "60000/60000 [==============================] - 8s 131us/step - loss: 0.0454 - val_loss: 0.0227\n",
            "Epoch 2/5\n",
            "60000/60000 [==============================] - 7s 117us/step - loss: 0.0174 - val_loss: 0.0127\n",
            "Epoch 3/5\n",
            "60000/60000 [==============================] - 7s 113us/step - loss: 0.0108 - val_loss: 0.0086\n",
            "Epoch 4/5\n",
            "60000/60000 [==============================] - 7s 114us/step - loss: 0.0078 - val_loss: 0.0066\n",
            "Epoch 5/5\n",
            "60000/60000 [==============================] - 7s 117us/step - loss: 0.0062 - val_loss: 0.0055\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "Ef3PZ2aqyHOe",
        "colab_type": "code",
        "outputId": "fc23dd9c-c707-4d20-997b-8ee5f2e2fe44",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 409
        }
      },
      "cell_type": "code",
      "source": [
        "conv_encoder = Model(x, h)\n",
        "encoded_imgs = conv_encoder.predict(X_test)\n",
        "\n",
        "n = 10\n",
        "plt.figure(figsize=(20, 8))\n",
        "for i in range(n):\n",
        "    ax = plt.subplot(1, n, i+1)\n",
        "    plt.imshow(encoded_imgs[i].reshape(4, 16).T)\n",
        "    plt.gray()\n",
        "    ax.get_xaxis().set_visible(False)\n",
        "    ax.get_yaxis().set_visible(False)\n",
        "plt.show()"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABGoAAAGICAYAAAAZAwYoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 1440x576 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "DV-LPOTyyMXw",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "decoded_imgs = autoencoder.predict(X_test)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "W50FUERzyYTD",
        "colab_type": "code",
        "outputId": "40248b95-6a17-4060-be15-8525ce1b5519",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 241
        }
      },
      "cell_type": "code",
      "source": [
        "n = 10\n",
        "plt.figure(figsize=(20, 6))\n",
        "for i in range(n):\n",
        "    # display original\n",
        "    ax = plt.subplot(3, n, i+1)\n",
        "    plt.imshow(X_test[i].reshape(28, 28))\n",
        "    plt.gray()\n",
        "    ax.get_xaxis().set_visible(False)\n",
        "    ax.get_yaxis().set_visible(False)\n",
        "\n",
        "    \n",
        "    # display reconstruction\n",
        "    ax = plt.subplot(3, n, i+n+1)\n",
        "    plt.imshow(decoded_imgs[i].reshape(28, 28))\n",
        "    plt.gray()\n",
        "    ax.get_xaxis().set_visible(False)\n",
        "    ax.get_yaxis().set_visible(False)\n",
        "    \n",
        "plt.show()"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABGoAAADgCAYAAABb04CrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAAByggc1AAAAAAAAOfH/ARmZY3NQS0AoAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1440x432 with 20 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}