Copy_of_caps2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of caps2.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/bklooste/tensorflowcollab/blob/master/Copy_of_caps2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "ecXSDaAn_xdD",
        "colab_type": "code",
        "outputId": "f61d9df1-365a-4ce9-9786-307c9fd115c9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import os\n",
        "import pandas as pd\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras import callbacks\n",
        "from keras.utils.vis_utils import plot_model\n",
        "\n",
        "import tensorflow as tf\n",
        "print(tf.__version__)"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1.13.0-rc1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "reTiD_HdAT3l",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import keras.backend as K\n",
        "import tensorflow as tf\n",
        "from keras import initializers, layers\n",
        "\n",
        "class Length(layers.Layer):\n",
        "    \"\"\"\n",
        "    Compute the length of vectors. This is used to compute a Tensor that has the same shape with y_true in margin_loss\n",
        "    inputs: shape=[dim_1, ..., dim_{n-1}, dim_n]\n",
        "    output: shape=[dim_1, ..., dim_{n-1}]\n",
        "    \"\"\"\n",
        "    def call(self, inputs, **kwargs):\n",
        "        return K.sqrt(K.sum(K.square(inputs), -1))\n",
        "\n",
        "    def compute_output_shape(self, input_shape):\n",
        "        return input_shape[:-1]\n",
        "\n",
        "class Mask(layers.Layer):\n",
        "    \"\"\"\n",
        "    Mask a Tensor with shape=[None, d1, d2] by the max value in axis=1.\n",
        "    Output shape: [None, d2]\n",
        "    \"\"\"\n",
        "    def call(self, inputs, **kwargs):\n",
        "        # use true label to select target capsule, shape=[batch_size, num_capsule]\n",
        "        if type(inputs) is list:  # true label is provided with shape = [batch_size, n_classes], i.e. one-hot code.\n",
        "            assert len(inputs) == 2\n",
        "            inputs, mask = inputs\n",
        "        else:  # if no true label, mask by the max length of vectors of capsules\n",
        "            x = inputs\n",
        "            # Enlarge the range of values in x to make max(new_x)=1 and others < 0\n",
        "            x = (x - K.max(x, 1, True)) / K.epsilon() + 1\n",
        "            mask = K.clip(x, 0, 1)  # the max value in x clipped to 1 and other to 0\n",
        "\n",
        "        # masked inputs, shape = [batch_size, dim_vector]\n",
        "        inputs_masked = K.batch_dot(inputs, mask, [1, 1])\n",
        "        return inputs_masked\n",
        "\n",
        "    def compute_output_shape(self, input_shape):\n",
        "        if type(input_shape[0]) is tuple:  # true label provided\n",
        "            return tuple([None, input_shape[0][-1]])\n",
        "        else:\n",
        "            return tuple([None, input_shape[-1]])\n",
        "\n",
        "\n",
        "def squash(vectors, axis=-1):\n",
        "    \"\"\"\n",
        "    The non-linear activation used in Capsule. It drives the length of a large vector to near 1 and small vector to 0\n",
        "    :param vectors: some vectors to be squashed, N-dim tensor\n",
        "    :param axis: the axis to squash\n",
        "    :return: a Tensor with same shape as input vectors\n",
        "    \"\"\"\n",
        "    s_squared_norm = K.sum(K.square(vectors), axis, keepdims=True)\n",
        "    scale = s_squared_norm / (1 + s_squared_norm) / K.sqrt(s_squared_norm)\n",
        "    return scale * vectors\n",
        "\n",
        "\n",
        "class CapsuleLayer(layers.Layer):\n",
        "    \"\"\"\n",
        "    The capsule layer. It is similar to Dense layer. Dense layer has `in_num` inputs, each is a scalar, the output of the \n",
        "    neuron from the former layer, and it has `out_num` output neurons. CapsuleLayer just expand the output of the neuron\n",
        "    from scalar to vector. So its input shape = [None, input_num_capsule, input_dim_vector] and output shape = \\\n",
        "    [None, num_capsule, dim_vector]. For Dense Layer, input_dim_vector = dim_vector = 1.\n",
        "    \n",
        "    :param num_capsule: number of capsules in this layer\n",
        "    :param dim_vector: dimension of the output vectors of the capsules in this layer\n",
        "    :param num_routings: number of iterations for the routing algorithm\n",
        "    \"\"\"\n",
        "    def __init__(self, num_capsule, dim_vector, num_routing=3,\n",
        "                 kernel_initializer='glorot_uniform',\n",
        "                 bias_initializer='zeros',\n",
        "                 **kwargs):\n",
        "        super(CapsuleLayer, self).__init__(**kwargs)\n",
        "        self.num_capsule = num_capsule\n",
        "        self.dim_vector = dim_vector\n",
        "        self.num_routing = num_routing\n",
        "        self.kernel_initializer = initializers.get(kernel_initializer)\n",
        "        self.bias_initializer = initializers.get(bias_initializer)\n",
        "\n",
        "    def build(self, input_shape):\n",
        "        assert len(input_shape) >= 3, \"The input Tensor should have shape=[None, input_num_capsule, input_dim_vector]\"\n",
        "        self.input_num_capsule = input_shape[1]\n",
        "        self.input_dim_vector = input_shape[2]\n",
        "\n",
        "        # Transform matrix\n",
        "        self.W = self.add_weight(shape=[self.input_num_capsule, self.num_capsule, self.input_dim_vector, self.dim_vector],\n",
        "                                 initializer=self.kernel_initializer,\n",
        "                                 name='W')\n",
        "\n",
        "        # Coupling coefficient. The redundant dimensions are just to facilitate subsequent matrix calculation.\n",
        "        self.bias = self.add_weight(shape=[1, self.input_num_capsule, self.num_capsule, 1, 1],\n",
        "                                    initializer=self.bias_initializer,\n",
        "                                    name='bias',\n",
        "                                    trainable=False)\n",
        "        self.built = True\n",
        "\n",
        "    def call(self, inputs, training=None):\n",
        "        # inputs.shape=[None, input_num_capsule, input_dim_vector]\n",
        "        # Expand dims to [None, input_num_capsule, 1, 1, input_dim_vector]\n",
        "        inputs_expand = K.expand_dims(K.expand_dims(inputs, 2), 2)\n",
        "\n",
        "        # Replicate num_capsule dimension to prepare being multiplied by W\n",
        "        # Now it has shape = [None, input_num_capsule, num_capsule, 1, input_dim_vector]\n",
        "        inputs_tiled = K.tile(inputs_expand, [1, 1, self.num_capsule, 1, 1])\n",
        "\n",
        "        \"\"\"  \n",
        "        # Compute `inputs * W` by expanding the first dim of W. More time-consuming and need batch_size.\n",
        "        # Now W has shape  = [batch_size, input_num_capsule, num_capsule, input_dim_vector, dim_vector]\n",
        "        w_tiled = K.tile(K.expand_dims(self.W, 0), [self.batch_size, 1, 1, 1, 1])\n",
        "        \n",
        "        # Transformed vectors, inputs_hat.shape = [None, input_num_capsule, num_capsule, 1, dim_vector]\n",
        "        inputs_hat = K.batch_dot(inputs_tiled, w_tiled, [4, 3])\n",
        "        \"\"\"\n",
        "        # Compute `inputs * W` by scanning inputs_tiled on dimension 0. This is faster but requires Tensorflow.\n",
        "        # inputs_hat.shape = [None, input_num_capsule, num_capsule, 1, dim_vector]\n",
        "        inputs_hat = tf.scan(lambda ac, x: K.batch_dot(x, self.W, [3, 2]),\n",
        "                             elems=inputs_tiled,\n",
        "                             initializer=K.zeros([self.input_num_capsule, self.num_capsule, 1, self.dim_vector]))\n",
        "        \"\"\"\n",
        "        # Routing algorithm V1. Use tf.while_loop in a dynamic way.\n",
        "        def body(i, b, outputs):\n",
        "            c = tf.nn.softmax(self.bias, dim=2)  # dim=2 is the num_capsule dimension\n",
        "            outputs = squash(K.sum(c * inputs_hat, 1, keepdims=True))\n",
        "            b = b + K.sum(inputs_hat * outputs, -1, keepdims=True)\n",
        "            return [i-1, b, outputs]\n",
        "\n",
        "        cond = lambda i, b, inputs_hat: i > 0\n",
        "        loop_vars = [K.constant(self.num_routing), self.bias, K.sum(inputs_hat, 1, keepdims=True)]\n",
        "        _, _, outputs = tf.while_loop(cond, body, loop_vars)\n",
        "        \"\"\"\n",
        "        # Routing algorithm V2. Use iteration. V2 and V1 both work without much difference on performance\n",
        "        assert self.num_routing > 0, 'The num_routing should be > 0.'\n",
        "        for i in range(self.num_routing):\n",
        "            c = tf.nn.softmax(self.bias, dim=2)  # dim=2 is the num_capsule dimension\n",
        "            # outputs.shape=[None, 1, num_capsule, 1, dim_vector]\n",
        "            outputs = squash(K.sum(c * inputs_hat, 1, keepdims=True))\n",
        "\n",
        "            # last iteration needs not compute bias which will not be passed to the graph any more anyway.\n",
        "            if i != self.num_routing - 1:\n",
        "                # self.bias = K.update_add(self.bias, K.sum(inputs_hat * outputs, [0, -1], keepdims=True))\n",
        "                self.bias += K.sum(inputs_hat * outputs, -1, keepdims=True)\n",
        "            # tf.summary.histogram('BigBee', self.bias)  # for debugging\n",
        "        return K.reshape(outputs, [-1, self.num_capsule, self.dim_vector])\n",
        "\n",
        "    def compute_output_shape(self, input_shape):\n",
        "        return tuple([None, self.num_capsule, self.dim_vector])\n",
        "\n",
        "\n",
        "def PrimaryCap(inputs, dim_vector, n_channels, kernel_size, strides, padding):\n",
        "    \"\"\"\n",
        "    Apply Conv2D `n_channels` times and concatenate all capsules\n",
        "    :param inputs: 4D tensor, shape=[None, width, height, channels]\n",
        "    :param dim_vector: the dim of the output vector of capsule\n",
        "    :param n_channels: the number of types of capsules\n",
        "    :return: output tensor, shape=[None, num_capsule, dim_vector]\n",
        "    \"\"\"\n",
        "    output = layers.Conv2D(filters=dim_vector*n_channels, kernel_size=kernel_size, strides=strides, padding=padding)(inputs)\n",
        "    outputs = layers.Reshape(target_shape=[-1, dim_vector])(output)\n",
        "    return layers.Lambda(squash)(outputs)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "H4sChQFYLnoX",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "0fpBXmI3Acy5",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras import layers, models\n",
        "from keras import backend as K\n",
        "from keras.utils import to_categorical\n",
        "def CapsNet(input_shape, n_class, num_routing):\n",
        "    \"\"\"\n",
        "    A Capsule Network on MNIST.\n",
        "    :param input_shape: data shape, 4d, [None, width, height, channels]\n",
        "    :param n_class: number of classes\n",
        "    :param num_routing: number of routing iterations\n",
        "    :return: A Keras Model with 2 inputs and 2 outputs\n",
        "    \"\"\"\n",
        "    x = layers.Input(shape=input_shape)\n",
        "\n",
        "    # Layer 1: Just a conventional Conv2D layer\n",
        "    conv1 = layers.Conv2D(filters=256, kernel_size=9, strides=1, padding='valid', activation='relu', name='conv1')(x)\n",
        "\n",
        "    # Layer 2: Conv2D layer with `squash` activation, then reshape to [None, num_capsule, dim_vector]\n",
        "    primarycaps = PrimaryCap(conv1, dim_vector=8, n_channels=32, kernel_size=9, strides=2, padding='valid')\n",
        "\n",
        "    # Layer 3: Capsule layer. Routing algorithm works here.\n",
        "    digitcaps = CapsuleLayer(num_capsule=n_class, dim_vector=16, num_routing=num_routing, name='digitcaps')(primarycaps)\n",
        "\n",
        "    # Layer 4: This is an auxiliary layer to replace each capsule with its length. Just to match the true label's shape.\n",
        "    # If using tensorflow, this will not be necessary. :)\n",
        "    out_caps = Length(name='out_caps')(digitcaps)\n",
        "\n",
        "    # Decoder network.\n",
        "    y = layers.Input(shape=(n_class,))\n",
        "    masked = Mask()([digitcaps, y])  # The true label is used to mask the output of capsule layer.\n",
        "    x_recon = layers.Dense(512, activation='relu')(masked)\n",
        "    x_recon = layers.Dense(1024, activation='relu')(x_recon)\n",
        "    x_recon = layers.Dense(784, activation='sigmoid')(x_recon)\n",
        "    x_recon = layers.Reshape(target_shape=[28, 28, 1], name='out_recon')(x_recon)\n",
        "\n",
        "    # two-input-two-output keras Model\n",
        "    return models.Model([x, y], [out_caps, x_recon])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ce8dylYJAf18",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def margin_loss(y_true, y_pred):\n",
        "    \"\"\"\n",
        "    Margin loss for Eq.(4). When y_true[i, :] contains not just one `1`, this loss should work too. Not test it.\n",
        "    :param y_true: [None, n_classes]\n",
        "    :param y_pred: [None, num_capsule]\n",
        "    :return: a scalar loss value.\n",
        "    \"\"\"\n",
        "    L = y_true * K.square(K.maximum(0., 0.9 - y_pred)) + \\\n",
        "        0.5 * (1 - y_true) * K.square(K.maximum(0., y_pred - 0.1))\n",
        "\n",
        "    return K.mean(K.sum(L, 1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "jnKfztXoAjk6",
        "colab_type": "code",
        "outputId": "c6a7048b-9397-44a4-f47e-aef67de487a9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 632
        }
      },
      "cell_type": "code",
      "source": [
        "# define model\n",
        "model = CapsNet(input_shape=[28, 28, 1],\n",
        "                n_class=10,\n",
        "                num_routing=3)\n",
        "model.summary()\n",
        "try:\n",
        "    plot_model(model, to_file='model.png', show_shapes=True)\n",
        "except Exception as e:\n",
        "    print('No fancy plot {}'.format(e))"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Variable += will be deprecated. Use variable.assign_add if you want assignment to the variable value or 'x = x + y' if you want a new python Tensor object.\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_3 (InputLayer)            (None, 28, 28, 1)    0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv1 (Conv2D)                  (None, 20, 20, 256)  20992       input_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_2 (Conv2D)               (None, 6, 6, 256)    5308672     conv1[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "reshape_2 (Reshape)             (None, 1152, 8)      0           conv2d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "lambda_2 (Lambda)               (None, 1152, 8)      0           reshape_2[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "digitcaps (CapsuleLayer)        (None, 10, 16)       1486080     lambda_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "input_4 (InputLayer)            (None, 10)           0                                            \n",
            "__________________________________________________________________________________________________\n",
            "mask_2 (Mask)                   (None, 16)           0           digitcaps[0][0]                  \n",
            "                                                                 input_4[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_4 (Dense)                 (None, 512)          8704        mask_2[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "dense_5 (Dense)                 (None, 1024)         525312      dense_4[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_6 (Dense)                 (None, 784)          803600      dense_5[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "out_caps (Length)               (None, 10)           0           digitcaps[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "out_recon (Reshape)             (None, 28, 28, 1)    0           dense_6[0][0]                    \n",
            "==================================================================================================\n",
            "Total params: 8,153,360\n",
            "Trainable params: 8,141,840\n",
            "Non-trainable params: 11,520\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "ON5-muO8AolH",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "data_train = pd.read_csv('https://raw.githubusercontent.com/bklooste/tensorflowcollab/master/capdata/train.csv')\n",
        "#data_train = pd.read_csv('../input/train.csv')\n",
        "X_full = data_train.iloc[:,1:]\n",
        "y_full = data_train.iloc[:,:1]\n",
        "x_train, x_test, y_train, y_test = train_test_split(X_full, y_full, test_size = 0.3)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "nMIRWUWOLbuI",
        "colab_type": "code",
        "outputId": "042fd237-6780-4505-9771-b9bb94d99e1e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 718
        }
      },
      "cell_type": "code",
      "source": [
        "#show data \n",
        "data_train.head(20)"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>label</th>\n",
              "      <th>pixel0</th>\n",
              "      <th>pixel1</th>\n",
              "      <th>pixel2</th>\n",
              "      <th>pixel3</th>\n",
              "      <th>pixel4</th>\n",
              "      <th>pixel5</th>\n",
              "      <th>pixel6</th>\n",
              "      <th>pixel7</th>\n",
              "      <th>pixel8</th>\n",
              "      <th>...</th>\n",
              "      <th>pixel774</th>\n",
              "      <th>pixel775</th>\n",
              "      <th>pixel776</th>\n",
              "      <th>pixel777</th>\n",
              "      <th>pixel778</th>\n",
              "      <th>pixel779</th>\n",
              "      <th>pixel780</th>\n",
              "      <th>pixel781</th>\n",
              "      <th>pixel782</th>\n",
              "      <th>pixel783</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>4</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>7</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>8</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>9</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>7</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>20 rows × 785 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "    label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \\\n",
              "0       1       0       0       0       0       0       0       0       0   \n",
              "1       0       0       0       0       0       0       0       0       0   \n",
              "2       1       0       0       0       0       0       0       0       0   \n",
              "3       4       0       0       0       0       0       0       0       0   \n",
              "4       0       0       0       0       0       0       0       0       0   \n",
              "5       0       0       0       0       0       0       0       0       0   \n",
              "6       7       0       0       0       0       0       0       0       0   \n",
              "7       3       0       0       0       0       0       0       0       0   \n",
              "8       5       0       0       0       0       0       0       0       0   \n",
              "9       3       0       0       0       0       0       0       0       0   \n",
              "10      8       0       0       0       0       0       0       0       0   \n",
              "11      9       0       0       0       0       0       0       0       0   \n",
              "12      1       0       0       0       0       0       0       0       0   \n",
              "13      3       0       0       0       0       0       0       0       0   \n",
              "14      3       0       0       0       0       0       0       0       0   \n",
              "15      1       0       0       0       0       0       0       0       0   \n",
              "16      2       0       0       0       0       0       0       0       0   \n",
              "17      0       0       0       0       0       0       0       0       0   \n",
              "18      7       0       0       0       0       0       0       0       0   \n",
              "19      5       0       0       0       0       0       0       0       0   \n",
              "\n",
              "    pixel8    ...     pixel774  pixel775  pixel776  pixel777  pixel778  \\\n",
              "0        0    ...            0         0         0         0         0   \n",
              "1        0    ...            0         0         0         0         0   \n",
              "2        0    ...            0         0         0         0         0   \n",
              "3        0    ...            0         0         0         0         0   \n",
              "4        0    ...            0         0         0         0         0   \n",
              "5        0    ...            0         0         0         0         0   \n",
              "6        0    ...            0         0         0         0         0   \n",
              "7        0    ...            0         0         0         0         0   \n",
              "8        0    ...            0         0         0         0         0   \n",
              "9        0    ...            0         0         0         0         0   \n",
              "10       0    ...            0         0         0         0         0   \n",
              "11       0    ...            0         0         0         0         0   \n",
              "12       0    ...            0         0         0         0         0   \n",
              "13       0    ...            0         0         0         0         0   \n",
              "14       0    ...            0         0         0         0         0   \n",
              "15       0    ...            0         0         0         0         0   \n",
              "16       0    ...            0         0         0         0         0   \n",
              "17       0    ...            0         0         0         0         0   \n",
              "18       0    ...            0         0         0         0         0   \n",
              "19       0    ...            0         0         0         0         0   \n",
              "\n",
              "    pixel779  pixel780  pixel781  pixel782  pixel783  \n",
              "0          0         0         0         0         0  \n",
              "1          0         0         0         0         0  \n",
              "2          0         0         0         0         0  \n",
              "3          0         0         0         0         0  \n",
              "4          0         0         0         0         0  \n",
              "5          0         0         0         0         0  \n",
              "6          0         0         0         0         0  \n",
              "7          0         0         0         0         0  \n",
              "8          0         0         0         0         0  \n",
              "9          0         0         0         0         0  \n",
              "10         0         0         0         0         0  \n",
              "11         0         0         0         0         0  \n",
              "12         0         0         0         0         0  \n",
              "13         0         0         0         0         0  \n",
              "14         0         0         0         0         0  \n",
              "15         0         0         0         0         0  \n",
              "16         0         0         0         0         0  \n",
              "17         0         0         0         0         0  \n",
              "18         0         0         0         0         0  \n",
              "19         0         0         0         0         0  \n",
              "\n",
              "[20 rows x 785 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "metadata": {
        "id": "tNs7EMF0AqZY",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "x_train = x_train.values.reshape(-1, 28, 28, 1).astype('float32') / 255.\n",
        "x_test = x_test.values.reshape(-1, 28, 28, 1).astype('float32') / 255.\n",
        "y_train = to_categorical(y_train.astype('float32'))\n",
        "y_test = to_categorical(y_test.astype('float32'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "GGs_qPk8AtoN",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def train(model, data, epoch_size_frac=1.0):\n",
        "    \"\"\"\n",
        "    Training a CapsuleNet\n",
        "    :param model: the CapsuleNet model\n",
        "    :param data: a tuple containing training and testing data, like `((x_train, y_train), (x_test, y_test))`\n",
        "    :param args: arguments\n",
        "    :return: The trained model\n",
        "    \"\"\"\n",
        "    # unpacking the data\n",
        "    (x_train, y_train), (x_test, y_test) = data\n",
        "\n",
        "    # callbacks\n",
        "    log = callbacks.CSVLogger('log.csv')\n",
        "    checkpoint = callbacks.ModelCheckpoint('weights-{epoch:02d}.h5',\n",
        "                                           save_best_only=True, save_weights_only=True, verbose=1)\n",
        "    lr_decay = callbacks.LearningRateScheduler(schedule=lambda epoch: 0.001 * np.exp(-epoch / 10.))\n",
        "\n",
        "    # compile the model\n",
        "    model.compile(optimizer='adam',\n",
        "                  loss=[margin_loss, 'mse'],\n",
        "                  loss_weights=[1., 0.0005],\n",
        "                  metrics={'out_caps': 'accuracy'})\n",
        "\n",
        "    \"\"\"\n",
        "    # Training without data augmentation:\n",
        "    model.fit([x_train, y_train], [y_train, x_train], batch_size=args.batch_size, epochs=args.epochs,\n",
        "              validation_data=[[x_test, y_test], [y_test, x_test]], callbacks=[log, tb, checkpoint])\n",
        "    \"\"\"\n",
        "\n",
        "    # -----------------------------------Begin: Training with data augmentation -----------------------------------#\n",
        "    def train_generator(x, y, batch_size, shift_fraction=0.):\n",
        "        train_datagen = ImageDataGenerator(width_shift_range=shift_fraction,\n",
        "                                           height_shift_range=shift_fraction)  # shift up to 2 pixel for MNIST\n",
        "        generator = train_datagen.flow(x, y, batch_size=batch_size)\n",
        "        while 1:\n",
        "            x_batch, y_batch = generator.next()\n",
        "            yield ([x_batch, y_batch], [y_batch, x_batch])\n",
        "\n",
        "    # Training with data augmentation. If shift_fraction=0., also no augmentation.\n",
        "    model.fit_generator(generator=train_generator(x_train, y_train, 100, 0.1),\n",
        "                        steps_per_epoch=int(epoch_size_frac*y_train.shape[0] / 64),\n",
        "                        epochs=5,\n",
        "                        validation_data=[[x_test, y_test], [y_test, x_test]],\n",
        "                        callbacks=[log, checkpoint, lr_decay])\n",
        "    # -----------------------------------End: Training with data augmentation -----------------------------------#\n",
        "\n",
        "    model.save_weights('trained_model.h5')\n",
        "    print('Trained model saved to \\'trained_model.h5\\'')\n",
        "\n",
        "    return model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "OWVxY-wTAwqb",
        "colab_type": "code",
        "outputId": "1d31cfea-455d-4539-bc49-a62856cdd7a4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 411
        }
      },
      "cell_type": "code",
      "source": [
        "train(model=model, data=((x_train, y_train), (x_test[:60], y_test[:60])), \n",
        "      epoch_size_frac = 1.0) # do 10% of an epoch (takes too long)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "459/459 [==============================] - 197s 429ms/step - loss: 0.1329 - out_caps_loss: 0.1328 - out_recon_loss: 0.1295 - out_caps_acc: 0.8638 - val_loss: 0.0202 - val_out_caps_loss: 0.0201 - val_out_recon_loss: 0.0642 - val_out_caps_acc: 0.9833\n",
            "\n",
            "Epoch 00001: val_loss improved from inf to 0.02016, saving model to weights-01.h5\n",
            "Epoch 2/5\n",
            "459/459 [==============================] - 194s 424ms/step - loss: 0.0316 - out_caps_loss: 0.0315 - out_recon_loss: 0.0611 - out_caps_acc: 0.9789 - val_loss: 0.0197 - val_out_caps_loss: 0.0196 - val_out_recon_loss: 0.0631 - val_out_caps_acc: 0.9833\n",
            "\n",
            "Epoch 00002: val_loss improved from 0.02016 to 0.01968, saving model to weights-02.h5\n",
            "Epoch 3/5\n",
            "459/459 [==============================] - 195s 425ms/step - loss: 0.0215 - out_caps_loss: 0.0215 - out_recon_loss: 0.0604 - out_caps_acc: 0.9853 - val_loss: 0.0128 - val_out_caps_loss: 0.0128 - val_out_recon_loss: 0.0621 - val_out_caps_acc: 0.9833\n",
            "\n",
            "Epoch 00003: val_loss improved from 0.01968 to 0.01280, saving model to weights-03.h5\n",
            "Epoch 4/5\n",
            "459/459 [==============================] - 194s 422ms/step - loss: 0.0175 - out_caps_loss: 0.0174 - out_recon_loss: 0.0598 - out_caps_acc: 0.9881 - val_loss: 0.0087 - val_out_caps_loss: 0.0087 - val_out_recon_loss: 0.0608 - val_out_caps_acc: 1.0000\n",
            "\n",
            "Epoch 00004: val_loss improved from 0.01280 to 0.00875, saving model to weights-04.h5\n",
            "Epoch 5/5\n",
            "459/459 [==============================] - 192s 418ms/step - loss: 0.0143 - out_caps_loss: 0.0143 - out_recon_loss: 0.0590 - out_caps_acc: 0.9903 - val_loss: 0.0158 - val_out_caps_loss: 0.0158 - val_out_recon_loss: 0.0599 - val_out_caps_acc: 0.9833\n",
            "\n",
            "Epoch 00005: val_loss did not improve from 0.00875\n",
            "Trained model saved to 'trained_model.h5'\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.engine.training.Model at 0x7f62b45b0c50>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "metadata": {
        "id": "qnayCsf7A1zR",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def combine_images(generated_images):\n",
        "    num = generated_images.shape[0]\n",
        "    width = int(np.sqrt(num))\n",
        "    height = int(np.ceil(float(num)/width))\n",
        "    shape = generated_images.shape[1:3]\n",
        "    image = np.zeros((height*shape[0], width*shape[1]),\n",
        "                     dtype=generated_images.dtype)\n",
        "    for index, img in enumerate(generated_images):\n",
        "        i = int(index/width)\n",
        "        j = index % width\n",
        "        image[i*shape[0]:(i+1)*shape[0], j*shape[1]:(j+1)*shape[1]] = \\\n",
        "            img[:, :, 0]\n",
        "    return image\n",
        "\n",
        "def test(model, data):\n",
        "    x_test, y_test = data\n",
        "    y_pred, x_recon = model.predict([x_test, y_test], batch_size=100)\n",
        "    print('-'*50)\n",
        "    print('Test acc:', np.sum(np.argmax(y_pred, 1) == np.argmax(y_test, 1))/y_test.shape[0])\n",
        "\n",
        "    import matplotlib.pyplot as plt\n",
        "    from PIL import Image\n",
        "\n",
        "    img = combine_images(np.concatenate([x_test[:50],x_recon[:50]]))\n",
        "    image = img * 255\n",
        "    Image.fromarray(image.astype(np.uint8)).save(\"real_and_recon.png\")\n",
        "    print()\n",
        "    print('Reconstructed images are saved to ./real_and_recon.png')\n",
        "    print('-'*50)\n",
        "    plt.imshow(plt.imread(\"real_and_recon.png\", ))\n",
        "    plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "HVg5HiMbA6Pn",
        "colab_type": "code",
        "outputId": "a91e3562-0507-43b3-d67e-172bbf38dd30",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 435
        }
      },
      "cell_type": "code",
      "source": [
        "test(model=model, data=(x_test[:100], y_test[:100]))"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--------------------------------------------------\n",
            "Test acc: 0.99\n",
            "\n",
            "Reconstructed images are saved to ./real_and_recon.png\n",
            "--------------------------------------------------\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVEAAAFNCAYAAAC5YlyiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "pdcGZJ2yBBEq",
        "colab_type": "code",
        "outputId": "1955bc54-f1e3-4016-bfc0-36453b943310",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "data_test = pd.read_csv('https://raw.githubusercontent.com/bklooste/tensorflowcollab/master/capdata/test.csv')\n",
        "#data_train = pd.read_csv('../input/train.csv')\n",
        "#data_test = pd.read_csv('../input/test.csv')\n",
        "data_test = data_test.values.reshape(-1, 28, 28, 1).astype('float32') / 255.\n",
        "y_pred, _ = model.predict([data_test, \n",
        "                           np.zeros((data_test.shape[0],10))], # empty values for the second vector \n",
        "                           batch_size = 32, verbose = True)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "28000/28000 [==============================] - 66s 2ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "2pxnvF69L8Id",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "with open('submission.csv', 'w') as out_file:\n",
        "    out_file.write('ImageId,Label\\n')\n",
        "    for img_id, guess_label in enumerate(np.argmax(y_pred,1),1):\n",
        "        out_file.write('%d,%d\\n' % (img_id, guess_label))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "sTkfaCDzSQvh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "557d59e6-2870-4669-f9a9-bfcc3182e418"
      },
      "cell_type": "code",
      "source": [
        "os.listdir('.')"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['.config',\n",
              " 'log.csv',\n",
              " 'weights-02.h5',\n",
              " 'real_and_recon.png',\n",
              " 'submission.csv',\n",
              " 'weights-03.h5',\n",
              " 'model.png',\n",
              " 'weights-04.h5',\n",
              " 'weights-01.h5',\n",
              " 'trained_model.h5',\n",
              " 'sample_data']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "metadata": {
        "id": "Ihd9FBzHS4Ks",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 197
        },
        "outputId": "6722e405-1292-419b-8c09-98f1f681decd"
      },
      "cell_type": "code",
      "source": [
        "from google.colab import files\n",
        "files.download('submission.csv')\n",
        "files.download('model.png')\n",
        "files.download('real_and_recon.png')\n"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "error",
          "ename": "TypeError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-34-43321458775b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mfiles\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'model.png'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mfiles\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'real_and_recon.png'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0mfiles\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;31mTypeError\u001b[0m: download() missing 1 required positional argument: 'filename'"
          ]
        }
      ]
    }
  ]
}