2_keras_mnist_advanced.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of keras_mnist_2.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/peterjsadowski/keras_tutorial/blob/master/2_keras_mnist_advanced.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RPbBpk1NzpQ1"
      },
      "source": [
        "# Advanced Keras Tutorial\n",
        "This tutorial introduces:\n",
        "\n",
        "1. Using the GPU\n",
        "2. Data generators in Keras.\n",
        "3. Functional API for complex models.\n",
        "\n",
        "The goal is to take two images of digits, and add them together. The model is never trained to recognize individual digits, but must learn this in order to solve the more complex problem.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B8zOyh7XzpQ3"
      },
      "source": [
        "import numpy as np\n",
        "import numpy.random\n",
        "import matplotlib.pyplot as plt\n",
        "import os, sys, socket\n",
        "\n",
        "import tensorflow as tf\n",
        "\n",
        "import keras\n",
        "from keras.datasets import mnist\n",
        "from keras.models import Model\n",
        "from keras.layers import Dense, Dropout, Activation, Input, Conv2D, Flatten\n",
        "from keras.layers import Concatenate\n",
        "from keras.optimizers import SGD, Adam"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lrm9TejgzpQ9"
      },
      "source": [
        "## Define data generators for training and validation."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vBOZOXe_zpQ-"
      },
      "source": [
        "# The following function is used to serve up both train and validation data.\n",
        "# Define a custom data generator class\n",
        "class CustomDataGenerator(keras.utils.PyDataset):\n",
        "    def __init__(self, X, Y, batchsize):\n",
        "        self.X = X\n",
        "        self.Y = Y\n",
        "        self.batchsize = batchsize\n",
        "\n",
        "    def __len__(self):\n",
        "        # Calculate the number of batches per epoch\n",
        "        return int(np.ceil(len(self.X) / self.batchsize))\n",
        "\n",
        "    def __getitem__(self, index):\n",
        "        # Generate data for a single batch\n",
        "        N = self.X.shape[0]\n",
        "        indices1 = np.random.randint(low=0, high=N, size=(self.batchsize,))\n",
        "        indices2 = np.random.randint(low=0, high=N, size=(self.batchsize,))\n",
        "\n",
        "        X1 = self.X[indices1,...].astype('float32') / 255.0\n",
        "        X2 = self.X[indices2,...].astype('float32') / 255.0\n",
        "        Y1 = self.Y[indices1]\n",
        "        Y2 = self.Y[indices2]\n",
        "        T  = (Y1 + Y2) # Sum of values.\n",
        "\n",
        "        X1 = np.expand_dims(X1, axis=1) # For conv with theano, shape=(batchsize, channels, row, col).\n",
        "        X2 = np.expand_dims(X2, axis=1) # We are just adding a dummy dimension saying that there is one channel.\n",
        "\n",
        "        Y1 = keras.utils.to_categorical(Y1, 10)\n",
        "        Y2 = keras.utils.to_categorical(Y2, 10)\n",
        "        T  = keras.utils.to_categorical(T, 19) # 19 possible values.\n",
        "\n",
        "        return {'input1':X1 , 'input2':X2},  {'out':T}\n",
        "\n",
        "\n",
        "\n",
        "def data_generator(X, Y, batchsize):\n",
        "    nb_classes = 10\n",
        "    N = X.shape[0]\n",
        "    while True:\n",
        "        indices1 = np.random.randint(low=0, high=N, size=(batchsize,))\n",
        "        indices2 = np.random.randint(low=0, high=N, size=(batchsize,))\n",
        "\n",
        "        X1 = X[indices1,...].astype('float32') / 255.0\n",
        "        X2 = X[indices2,...].astype('float32') / 255.0\n",
        "        Y1 = Y[indices1]\n",
        "        Y2 = Y[indices2]\n",
        "        T  = (Y1 + Y2) # Sum of values.\n",
        "\n",
        "        X1 = np.expand_dims(X1, axis=1) # For conv with theano, shape=(batchsize, channels, row, col).\n",
        "        X2 = np.expand_dims(X2, axis=1) # We are just adding a dummy dimension saying that there is one channel.\n",
        "\n",
        "        Y1 = keras.utils.to_categorical(Y1, nb_classes)\n",
        "        Y2 = keras.utils.to_categorical(Y2, nb_classes)\n",
        "        T  = keras.utils.to_categorical(T, 19) # 19 possible values.\n",
        "\n",
        "        yield {'input1':X1 , 'input2':X2}, T #  {'out':T} #, 'aux1':Y1, 'aux2':Y2}\n",
        "\n",
        "# Load data.\n",
        "(X_train, Y_train), (X_valid, Y_valid) = mnist.load_data() # Shape = (N,28,28), (N,)\n",
        "# Create generators.\n",
        "batchsize  = 200\n",
        "data_train = CustomDataGenerator(X_train, Y_train, batchsize)\n",
        "data_valid = CustomDataGenerator(X_valid, Y_valid, batchsize)"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gui8KjR-zpRB",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 213
        },
        "outputId": "6495d52a-ba7c-4da7-e4f3-541668bbba07"
      },
      "source": [
        "# Plot examples of the data.\n",
        "inputs, outputs = data_train.__getitem__(index=0)\n",
        "Xbatch = inputs['input1']\n",
        "plt.figure(1, figsize=(14,2))\n",
        "for i in range(10):\n",
        "    plt.subplot(2,10,i+1)\n",
        "    plt.imshow(inputs['input1'][i].reshape(28,28), cmap='gray', interpolation='nearest')\n",
        "    plt.xticks([])\n",
        "    plt.yticks([])\n",
        "    plt.title('Target=%d' % np.argmax(outputs['out'][i]))\n",
        "    plt.subplot(2,10,10+i+1)\n",
        "    plt.imshow(inputs['input2'][i].reshape(28,28), cmap='gray', interpolation='nearest')\n",
        "    plt.xticks([])\n",
        "    plt.yticks([])"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1400x200 with 20 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rNfItrsTzpRF"
      },
      "source": [
        "# Define Keras Model with the Functional API\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dGg_p1h2zpRH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 481
        },
        "outputId": "814584c0-d4b8-4c05-aee4-4a8c01e1cb3b"
      },
      "source": [
        "# Input data.\n",
        "input1   = Input(shape=(1, 28,28), dtype='float32', name='input1') # Argument 'name' must match name in dictionary.\n",
        "input2   = Input(shape=(1, 28,28), dtype='float32', name='input2')\n",
        "\n",
        "# Layer one: Single convolutional layer applied to each input independently, with \"shared weights.\"\n",
        "nb_filter      = 32     # Number of convolutional kernels.\n",
        "nb_row, nb_col = 7,7   # Convolution kernel size.\n",
        "strides        = (3,3) # Step size for convolution kernels.\n",
        "conv  =  Conv2D(nb_filter, (nb_row, nb_col), activation='relu', padding='same', strides=strides)\n",
        "x1    = conv(input1) # Layer object conv transforms data.\n",
        "x2    = conv(input2)\n",
        "\n",
        "# Flatten data: transform from (28,28) to (784,)\n",
        "x1 = Flatten()(x1)\n",
        "x2 = Flatten()(x2)\n",
        "\n",
        "# Layer two: Single fully-connected layer applied to each input independently, with shared weights.\n",
        "layer = Dense(256, activation='relu')\n",
        "x1 = layer(x1)\n",
        "x2 = layer(x2)\n",
        "\n",
        "# Auxiliary softmax layers try to classify digits.\n",
        "# The output of these layers aren't fed into the next layer.\n",
        "#layer = Dense(10) # These weights are shared.\n",
        "#aux1  = Activation(activation='softmax', name='aux1')(layer(x1)) # Output layers must be named.\n",
        "#aux2  = Activation(activation='softmax', name='aux2')(layer(x2))\n",
        "\n",
        "# Merge hidden representations.\n",
        "#x = merge.Concatenate()([x1, x2])\n",
        "x = Concatenate()([x1, x2])\n",
        "\n",
        "# More dense layers then output.\n",
        "x   = Dense(512, activation='relu')(x)\n",
        "out = Dense(19, activation='softmax', name='out')(x) # Output layers must be named.\n",
        "\n",
        "# Create model object that puts it all together.\n",
        "#model = Model(inputs=[input1, input2], outputs=[out, aux1, aux2])\n",
        "model = Model(inputs=[input1, input2], outputs=[out])\n",
        "model.summary()"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1mModel: \"functional\"\u001b[0m\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\">Model: \"functional\"</span>\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┓\n",
              "┃\u001b[1m \u001b[0m\u001b[1mLayer (type)             \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape          \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m       Param #\u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mConnected to          \u001b[0m\u001b[1m \u001b[0m┃\n",
              "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━┩\n",
              "│ input1 (\u001b[38;5;33mInputLayer\u001b[0m)       │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m1\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m28\u001b[0m)      │              \u001b[38;5;34m0\u001b[0m │ -                      │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ input2 (\u001b[38;5;33mInputLayer\u001b[0m)       │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m1\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m28\u001b[0m)      │              \u001b[38;5;34m0\u001b[0m │ -                      │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ conv2d (\u001b[38;5;33mConv2D\u001b[0m)           │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m1\u001b[0m, \u001b[38;5;34m10\u001b[0m, \u001b[38;5;34m32\u001b[0m)      │         \u001b[38;5;34m43,936\u001b[0m │ input1[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m],          │\n",
              "│                           │                        │                │ input2[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m]           │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ flatten (\u001b[38;5;33mFlatten\u001b[0m)         │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m320\u001b[0m)            │              \u001b[38;5;34m0\u001b[0m │ conv2d[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m]           │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ flatten_1 (\u001b[38;5;33mFlatten\u001b[0m)       │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m320\u001b[0m)            │              \u001b[38;5;34m0\u001b[0m │ conv2d[\u001b[38;5;34m1\u001b[0m][\u001b[38;5;34m0\u001b[0m]           │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ dense (\u001b[38;5;33mDense\u001b[0m)             │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m256\u001b[0m)            │         \u001b[38;5;34m82,176\u001b[0m │ flatten[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m],         │\n",
              "│                           │                        │                │ flatten_1[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m]        │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ concatenate (\u001b[38;5;33mConcatenate\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m512\u001b[0m)            │              \u001b[38;5;34m0\u001b[0m │ dense[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m],           │\n",
              "│                           │                        │                │ dense[\u001b[38;5;34m1\u001b[0m][\u001b[38;5;34m0\u001b[0m]            │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ dense_1 (\u001b[38;5;33mDense\u001b[0m)           │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m512\u001b[0m)            │        \u001b[38;5;34m262,656\u001b[0m │ concatenate[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m]      │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ out (\u001b[38;5;33mDense\u001b[0m)               │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m19\u001b[0m)             │          \u001b[38;5;34m9,747\u001b[0m │ dense_1[\u001b[38;5;34m0\u001b[0m][\u001b[38;5;34m0\u001b[0m]          │\n",
              "└───────────────────────────┴────────────────────────┴────────────────┴────────────────────────┘\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\">┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┓\n",
              "┃<span style=\"font-weight: bold\"> Layer (type)              </span>┃<span style=\"font-weight: bold\"> Output Shape           </span>┃<span style=\"font-weight: bold\">        Param # </span>┃<span style=\"font-weight: bold\"> Connected to           </span>┃\n",
              "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━┩\n",
              "│ input1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">InputLayer</span>)       │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">28</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">28</span>)      │              <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │ -                      │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ input2 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">InputLayer</span>)       │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">28</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">28</span>)      │              <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │ -                      │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ conv2d (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv2D</span>)           │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">10</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">32</span>)      │         <span style=\"color: #00af00; text-decoration-color: #00af00\">43,936</span> │ input1[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>],          │\n",
              "│                           │                        │                │ input2[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]           │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ flatten (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Flatten</span>)         │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">320</span>)            │              <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │ conv2d[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]           │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ flatten_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Flatten</span>)       │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">320</span>)            │              <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │ conv2d[<span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]           │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ dense (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)             │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">256</span>)            │         <span style=\"color: #00af00; text-decoration-color: #00af00\">82,176</span> │ flatten[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>],         │\n",
              "│                           │                        │                │ flatten_1[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]        │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ concatenate (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Concatenate</span>) │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">512</span>)            │              <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │ dense[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>],           │\n",
              "│                           │                        │                │ dense[<span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]            │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ dense_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)           │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">512</span>)            │        <span style=\"color: #00af00; text-decoration-color: #00af00\">262,656</span> │ concatenate[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]      │\n",
              "├───────────────────────────┼────────────────────────┼────────────────┼────────────────────────┤\n",
              "│ out (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)               │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">19</span>)             │          <span style=\"color: #00af00; text-decoration-color: #00af00\">9,747</span> │ dense_1[<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>][<span style=\"color: #00af00; text-decoration-color: #00af00\">0</span>]          │\n",
              "└───────────────────────────┴────────────────────────┴────────────────┴────────────────────────┘\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1m Total params: \u001b[0m\u001b[38;5;34m398,515\u001b[0m (1.52 MB)\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Total params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">398,515</span> (1.52 MB)\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m398,515\u001b[0m (1.52 MB)\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Trainable params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">398,515</span> (1.52 MB)\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m0\u001b[0m (0.00 B)\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Non-trainable params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> (0.00 B)\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nXH5jZGHzpRL"
      },
      "source": [
        "# Optimization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j_KuS3jSzpRM"
      },
      "source": [
        "optimizer = Adam(learning_rate=0.001, beta_1=0.99, beta_2=0.999, epsilon=1e-08) # Optimization hyperparameters.\n",
        "\n",
        "model.compile(optimizer=optimizer,\n",
        "              loss={'out':'categorical_crossentropy'},\n",
        "              loss_weights=1.0, #{'out': 1.0}, #\n",
        "              metrics=['accuracy'])\n"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HMuCALu9zpRP"
      },
      "source": [
        "## Train the model."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hkgjaFS-zpRQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "91b7f7ce-4fac-46ef-aaa7-70bfa57cd582"
      },
      "source": [
        "# Callbacks can be used to stop early, decrease learning rate, checkpoint the model, etc.\n",
        "#from keras.callbacks import EarlyStopping\n",
        "#stopping  = keras.callbacks.EarlyStopping(monitor='val_loss', patience=0, verbose=0, mode='auto')\n",
        "callbacks = []#[stopping]\n",
        "\n",
        "# The fit_generator function loads data batches on the fly, instead of transfering entire data set to the gpu.\n",
        "history   = model.fit(x=data_train,\n",
        "                      steps_per_epoch=100, # 500\n",
        "                      epochs=10, verbose=1,\n",
        "                      callbacks=callbacks,\n",
        "                      validation_data=data_valid, validation_steps=100)"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/keras/src/trainers/data_adapters/py_dataset_adapter.py:121: UserWarning: Your `PyDataset` class should call `super().__init__(**kwargs)` in its constructor. `**kwargs` can include `workers`, `use_multiprocessing`, `max_queue_size`. Do not pass these arguments to `fit()`, as they will be ignored.\n",
            "  self._warn_if_super_not_called()\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m5s\u001b[0m 10ms/step - accuracy: 0.1196 - loss: 2.6684 - val_accuracy: 0.2550 - val_loss: 2.0819\n",
            "Epoch 2/10\n",
            "\u001b[1m 47/100\u001b[0m \u001b[32m━━━━━━━━━\u001b[0m\u001b[37m━━━━━━━━━━━\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.2727 - loss: 2.0343"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/lib/python3.10/contextlib.py:153: UserWarning: Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches. You may need to use the `.repeat()` function when building your dataset.\n",
            "  self.gen.throw(typ, value, traceback)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 5ms/step - accuracy: 0.3082 - loss: 1.9477 - val_accuracy: 0.5294 - val_loss: 1.4334\n",
            "Epoch 3/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6009 - loss: 1.2495 - val_accuracy: 0.7727 - val_loss: 0.7785\n",
            "Epoch 4/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.7721 - val_loss: 0.7798\n",
            "Epoch 5/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 6ms/step - accuracy: 0.7995 - loss: 0.6946 - val_accuracy: 0.8573 - val_loss: 0.4979\n",
            "Epoch 6/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8605 - loss: 0.4823 - val_accuracy: 0.8952 - val_loss: 0.3597\n",
            "Epoch 7/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.8971 - loss: 0.3610 - val_accuracy: 0.9177 - val_loss: 0.2779\n",
            "Epoch 8/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.9169 - val_loss: 0.2795\n",
            "Epoch 9/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 5ms/step - accuracy: 0.9247 - loss: 0.2601 - val_accuracy: 0.9293 - val_loss: 0.2309\n",
            "Epoch 10/10\n",
            "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.9394 - loss: 0.2061 - val_accuracy: 0.9366 - val_loss: 0.2136\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Vgj-AWt_zpRU",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 482
        },
        "outputId": "29e2475a-dfd7-4b81-8280-d42b6eb8ffaa"
      },
      "source": [
        "# Plot loss trajectory throughout training.\n",
        "plt.figure(1, figsize=(10,5))\n",
        "plt.plot(history.history['accuracy'], label='primary')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend()"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x79b276c21e40>"
            ]
          },
          "metadata": {},
          "execution_count": 35
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x500 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L6doGuvdzpRY"
      },
      "source": [
        "# Post-training analysis."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s1hFaHsOzpRZ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 341
        },
        "outputId": "01a93292-e041-4f8c-b0f2-4a2fe67902cb"
      },
      "source": [
        "# Plot example pairs for which the network made a mistake.\n",
        "inputs, outputs = data_valid.__getitem__(index=0)\n",
        "plt.figure(1, figsize=(14,3))\n",
        "i = 0\n",
        "while i < 10:\n",
        "    inputs, outputs = data_valid.__getitem__(index=0)\n",
        "    predictions = model.predict(inputs)\n",
        "    j = 0\n",
        "    while j < 100:\n",
        "        yhat   = predictions[j].argmax(axis=-1) # predictions is list: [out, aux1, aux2]\n",
        "        y      = np.argmax(outputs['out'][j])\n",
        "        input1 = inputs['input1'][j].reshape(28,28)\n",
        "        input2 = inputs['input2'][j].reshape(28,28)\n",
        "        j += 1\n",
        "        if yhat == y:\n",
        "            # Prediction was correct. Ignore this example.\n",
        "            continue\n",
        "        else:\n",
        "            # Prediction was incorrect. Show example.\n",
        "            plt.subplot(2,10,i+1)\n",
        "            plt.imshow(input1, cmap='gray', interpolation='nearest')\n",
        "            plt.xticks([])\n",
        "            plt.yticks([])\n",
        "            plt.title('Pred=%d' % yhat)\n",
        "            plt.title('$y=%d$\\n$\\hat{y}=%d$' % (y,yhat))\n",
        "            plt.subplot(2,10,10+i+1)\n",
        "            plt.imshow(input2, cmap='gray', interpolation='nearest')\n",
        "            plt.xticks([])\n",
        "            plt.yticks([])\n",
        "            i += 1\n",
        "        if not i<10:\n",
        "            break\n"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1m7/7\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step \n",
            "\u001b[1m7/7\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step \n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1400x300 with 20 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "rMxzznIKYOCO"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}