Caps_Net(Num_r=2).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.5"
    },
    "colab": {
      "name": "Caps-Net(Num_r=2).ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/HB5101/CapsNet_Project/blob/master/Caps_Net(Num_r%3D2).ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oZCdoCYqZA6F",
        "colab_type": "text"
      },
      "source": [
        "This is Keras implementation of CapsNet as per Geoffrey Hinton's paper Dynamic Routing Between Capsules using 2 routing iterations.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cvhVix4ASwJ1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from keras import layers,Sequential\n",
        "import keras.backend as K\n",
        "from keras import layers, Model, initializers\n",
        "from keras.layers import Input,Conv2D,Concatenate,Lambda,Reshape,Dense,Flatten,Reshape"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2K7HkQFUNbnD",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from layers_capsnet_py import *\n",
        "from datapreprocess import *"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-WzqMABwwO8i",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class Length(layers.Layer):\n",
        "    \"\"\"\n",
        "     To compute the length of input vectors. \n",
        "    inputs: shape=[None, num_vectors, dim_vector]\n",
        "    output: shape=[None, num_vectors]\n",
        "    \"\"\"\n",
        "    def call(self, inputs, **kwargs):\n",
        "        return K.sqrt(K.sum(K.square(inputs), -1) + K.epsilon())\n",
        "\n",
        "    def compute_output_shape(self, input_shape):\n",
        "        return input_shape[:-1]\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4B6oDkZRSwKY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class Mask(layers.Layer):\n",
        "   \n",
        "    def call(self, inputs, **kwargs):\n",
        "        if ((type(inputs) is list) and (len(inputs)==2)):  # true label is provided with shape = [None, n_classes], i.e. one-hot code.\n",
        "            inputs, mask = inputs\n",
        "        else:  # if no true label, mask by the max length of capsules,mainly used for prediction\n",
        "            # compute lengths of capsules\n",
        "            x = K.sqrt(K.sum(K.square(inputs), -1))\n",
        "            # mask.shape=[None, n_classes]=[None, num_capsule]\n",
        "            mask = K.one_hot(indices=K.argmax(x, 1), num_classes=x.get_shape().as_list()[1])\n",
        "\n",
        "        # inputs.shape=[None, num_capsule, dim_capsule]\n",
        "        # masked.shape=[None, num_capsule * dim_capsule]\n",
        "        masked =(inputs * K.expand_dims(mask, -1))\n",
        "        return masked\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WlsjrE-pSwKJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def CapsNet(input_shape, routings, n_class):\n",
        "\n",
        "    x= layers.Input(shape=input_shape)    #inputNode\n",
        "\n",
        "    y = layers.Input(shape=(n_class,))\n",
        "    \n",
        "    conv_layer = layers.Conv2D(filters=256, kernel_size=9, strides=1, padding='valid', activation='relu', name='conv_layer')(x) #ReluConv1\n",
        "    \n",
        "    primary_layer= PrimaryCap( conv_layer, dim_capsule=8, n_channels=32, kernel_size=9, strides=2, padding='valid')\n",
        "    \n",
        "    digitcapsule=DigiCap(num_capsule=n_class, dim_capsule=16, routings=routings, name='digitcapsule')(primary_layer)\n",
        "\n",
        "    # The length of the capsule's output vector \n",
        "    encoder = Length(name=\"encoder\")(digitcapsule)\n",
        "\n",
        "    # masking\n",
        "    masky = Mask()([digitcapsule, y])  \n",
        "    maskeval = Mask()(digitcapsule)\n",
        "    maskedy=Flatten()(masky)\n",
        "    masked=Flatten()(maskeval)\n",
        "\n",
        "    #decoder\n",
        "    decoder =Sequential(name='decoder')\n",
        "    decoder.add(layers.Dense(512, activation='relu', input_dim=16*n_class))\n",
        "    decoder.add(layers.Dense(1024, activation='relu'))\n",
        "    decoder.add(layers.Dense(784, activation='sigmoid'))\n",
        "    decoder.add(layers.Reshape(target_shape=input_shape))\n",
        "    \n",
        "    # two-input-two-output keras Model\n",
        "\n",
        "    #train_model\n",
        "    train_model=Model([x, y], [encoder, decoder(maskedy)])\n",
        "\n",
        "    #evalute_model\n",
        "    eval_model=Model(x, [encoder, decoder(masked)])\n",
        "   \n",
        "    \n",
        "    return train_model,eval_model\n",
        "   "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KuDMV5PeSwKM",
        "colab_type": "code",
        "outputId": "525aa9f1-37bc-4419-c36e-0e6e744ae396",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 138
        }
      },
      "source": [
        "train,evalm=CapsNet(input_shape=x_test.shape[1:],routings=2,n_class=10)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ByDEO-HKSwKO",
        "colab_type": "code",
        "outputId": "6c477fd3-2b1e-4f9b-9f36-132882f1f122",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "train.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_3 (InputLayer)            (None, 28, 28, 1)    0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv_layer (Conv2D)             (None, 20, 20, 256)  20992       input_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_33 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_34 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_35 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_36 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_37 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_38 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_39 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_40 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_41 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_42 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_43 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_44 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_45 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_46 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_47 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_48 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_49 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_50 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_51 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_52 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_53 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_54 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_55 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_56 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_57 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_58 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_59 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_60 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_61 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_62 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_63 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_64 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "reshape_34 (Reshape)            (None, Dimension(36) 0           conv2d_33[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_35 (Reshape)            (None, Dimension(36) 0           conv2d_34[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_36 (Reshape)            (None, Dimension(36) 0           conv2d_35[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_37 (Reshape)            (None, Dimension(36) 0           conv2d_36[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_38 (Reshape)            (None, Dimension(36) 0           conv2d_37[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_39 (Reshape)            (None, Dimension(36) 0           conv2d_38[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_40 (Reshape)            (None, Dimension(36) 0           conv2d_39[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_41 (Reshape)            (None, Dimension(36) 0           conv2d_40[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_42 (Reshape)            (None, Dimension(36) 0           conv2d_41[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_43 (Reshape)            (None, Dimension(36) 0           conv2d_42[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_44 (Reshape)            (None, Dimension(36) 0           conv2d_43[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_45 (Reshape)            (None, Dimension(36) 0           conv2d_44[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_46 (Reshape)            (None, Dimension(36) 0           conv2d_45[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_47 (Reshape)            (None, Dimension(36) 0           conv2d_46[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_48 (Reshape)            (None, Dimension(36) 0           conv2d_47[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_49 (Reshape)            (None, Dimension(36) 0           conv2d_48[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_50 (Reshape)            (None, Dimension(36) 0           conv2d_49[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_51 (Reshape)            (None, Dimension(36) 0           conv2d_50[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_52 (Reshape)            (None, Dimension(36) 0           conv2d_51[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_53 (Reshape)            (None, Dimension(36) 0           conv2d_52[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_54 (Reshape)            (None, Dimension(36) 0           conv2d_53[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_55 (Reshape)            (None, Dimension(36) 0           conv2d_54[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_56 (Reshape)            (None, Dimension(36) 0           conv2d_55[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_57 (Reshape)            (None, Dimension(36) 0           conv2d_56[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_58 (Reshape)            (None, Dimension(36) 0           conv2d_57[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_59 (Reshape)            (None, Dimension(36) 0           conv2d_58[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_60 (Reshape)            (None, Dimension(36) 0           conv2d_59[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_61 (Reshape)            (None, Dimension(36) 0           conv2d_60[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_62 (Reshape)            (None, Dimension(36) 0           conv2d_61[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_63 (Reshape)            (None, Dimension(36) 0           conv2d_62[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_64 (Reshape)            (None, Dimension(36) 0           conv2d_63[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_65 (Reshape)            (None, Dimension(36) 0           conv2d_64[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_2 (Concatenate)     (None, Dimension(115 0           reshape_34[0][0]                 \n",
            "                                                                 reshape_35[0][0]                 \n",
            "                                                                 reshape_36[0][0]                 \n",
            "                                                                 reshape_37[0][0]                 \n",
            "                                                                 reshape_38[0][0]                 \n",
            "                                                                 reshape_39[0][0]                 \n",
            "                                                                 reshape_40[0][0]                 \n",
            "                                                                 reshape_41[0][0]                 \n",
            "                                                                 reshape_42[0][0]                 \n",
            "                                                                 reshape_43[0][0]                 \n",
            "                                                                 reshape_44[0][0]                 \n",
            "                                                                 reshape_45[0][0]                 \n",
            "                                                                 reshape_46[0][0]                 \n",
            "                                                                 reshape_47[0][0]                 \n",
            "                                                                 reshape_48[0][0]                 \n",
            "                                                                 reshape_49[0][0]                 \n",
            "                                                                 reshape_50[0][0]                 \n",
            "                                                                 reshape_51[0][0]                 \n",
            "                                                                 reshape_52[0][0]                 \n",
            "                                                                 reshape_53[0][0]                 \n",
            "                                                                 reshape_54[0][0]                 \n",
            "                                                                 reshape_55[0][0]                 \n",
            "                                                                 reshape_56[0][0]                 \n",
            "                                                                 reshape_57[0][0]                 \n",
            "                                                                 reshape_58[0][0]                 \n",
            "                                                                 reshape_59[0][0]                 \n",
            "                                                                 reshape_60[0][0]                 \n",
            "                                                                 reshape_61[0][0]                 \n",
            "                                                                 reshape_62[0][0]                 \n",
            "                                                                 reshape_63[0][0]                 \n",
            "                                                                 reshape_64[0][0]                 \n",
            "                                                                 reshape_65[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "lambda_2 (Lambda)               (None, 1152, 8)      0           concatenate_2[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "digitcapsule (CapsuleLayer)     (None, 1152, 8)      1474560     lambda_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "input_4 (InputLayer)            (None, 10)           0                                            \n",
            "__________________________________________________________________________________________________\n",
            "mask_3 (Mask)                   [(None, 1152, 8), (N 0           digitcapsule[0][0]               \n",
            "                                                                 input_4[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "flatten_3 (Flatten)             (None, 9216)         0           mask_3[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "encoder (Length)                (None, 1152)         0           digitcapsule[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "decoder (Sequential)            (None, 28, 28, 1)    1411344     flatten_3[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 8,215,568\n",
            "Trainable params: 8,215,568\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RTyZJ5DuSwKP",
        "colab_type": "code",
        "outputId": "7803a85a-7851-4039-991a-d263330f1df5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "evalm.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_3 (InputLayer)            (None, 28, 28, 1)    0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv_layer (Conv2D)             (None, 20, 20, 256)  20992       input_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_33 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_34 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_35 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_36 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_37 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_38 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_39 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_40 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_41 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_42 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_43 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_44 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_45 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_46 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_47 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_48 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_49 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_50 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_51 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_52 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_53 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_54 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_55 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_56 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_57 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_58 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_59 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_60 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_61 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_62 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_63 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_64 (Conv2D)              (None, 6, 6, 8)      165896      conv_layer[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "reshape_34 (Reshape)            (None, Dimension(36) 0           conv2d_33[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_35 (Reshape)            (None, Dimension(36) 0           conv2d_34[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_36 (Reshape)            (None, Dimension(36) 0           conv2d_35[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_37 (Reshape)            (None, Dimension(36) 0           conv2d_36[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_38 (Reshape)            (None, Dimension(36) 0           conv2d_37[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_39 (Reshape)            (None, Dimension(36) 0           conv2d_38[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_40 (Reshape)            (None, Dimension(36) 0           conv2d_39[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_41 (Reshape)            (None, Dimension(36) 0           conv2d_40[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_42 (Reshape)            (None, Dimension(36) 0           conv2d_41[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_43 (Reshape)            (None, Dimension(36) 0           conv2d_42[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_44 (Reshape)            (None, Dimension(36) 0           conv2d_43[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_45 (Reshape)            (None, Dimension(36) 0           conv2d_44[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_46 (Reshape)            (None, Dimension(36) 0           conv2d_45[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_47 (Reshape)            (None, Dimension(36) 0           conv2d_46[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_48 (Reshape)            (None, Dimension(36) 0           conv2d_47[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_49 (Reshape)            (None, Dimension(36) 0           conv2d_48[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_50 (Reshape)            (None, Dimension(36) 0           conv2d_49[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_51 (Reshape)            (None, Dimension(36) 0           conv2d_50[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_52 (Reshape)            (None, Dimension(36) 0           conv2d_51[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_53 (Reshape)            (None, Dimension(36) 0           conv2d_52[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_54 (Reshape)            (None, Dimension(36) 0           conv2d_53[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_55 (Reshape)            (None, Dimension(36) 0           conv2d_54[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_56 (Reshape)            (None, Dimension(36) 0           conv2d_55[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_57 (Reshape)            (None, Dimension(36) 0           conv2d_56[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_58 (Reshape)            (None, Dimension(36) 0           conv2d_57[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_59 (Reshape)            (None, Dimension(36) 0           conv2d_58[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_60 (Reshape)            (None, Dimension(36) 0           conv2d_59[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_61 (Reshape)            (None, Dimension(36) 0           conv2d_60[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_62 (Reshape)            (None, Dimension(36) 0           conv2d_61[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_63 (Reshape)            (None, Dimension(36) 0           conv2d_62[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_64 (Reshape)            (None, Dimension(36) 0           conv2d_63[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_65 (Reshape)            (None, Dimension(36) 0           conv2d_64[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_2 (Concatenate)     (None, Dimension(115 0           reshape_34[0][0]                 \n",
            "                                                                 reshape_35[0][0]                 \n",
            "                                                                 reshape_36[0][0]                 \n",
            "                                                                 reshape_37[0][0]                 \n",
            "                                                                 reshape_38[0][0]                 \n",
            "                                                                 reshape_39[0][0]                 \n",
            "                                                                 reshape_40[0][0]                 \n",
            "                                                                 reshape_41[0][0]                 \n",
            "                                                                 reshape_42[0][0]                 \n",
            "                                                                 reshape_43[0][0]                 \n",
            "                                                                 reshape_44[0][0]                 \n",
            "                                                                 reshape_45[0][0]                 \n",
            "                                                                 reshape_46[0][0]                 \n",
            "                                                                 reshape_47[0][0]                 \n",
            "                                                                 reshape_48[0][0]                 \n",
            "                                                                 reshape_49[0][0]                 \n",
            "                                                                 reshape_50[0][0]                 \n",
            "                                                                 reshape_51[0][0]                 \n",
            "                                                                 reshape_52[0][0]                 \n",
            "                                                                 reshape_53[0][0]                 \n",
            "                                                                 reshape_54[0][0]                 \n",
            "                                                                 reshape_55[0][0]                 \n",
            "                                                                 reshape_56[0][0]                 \n",
            "                                                                 reshape_57[0][0]                 \n",
            "                                                                 reshape_58[0][0]                 \n",
            "                                                                 reshape_59[0][0]                 \n",
            "                                                                 reshape_60[0][0]                 \n",
            "                                                                 reshape_61[0][0]                 \n",
            "                                                                 reshape_62[0][0]                 \n",
            "                                                                 reshape_63[0][0]                 \n",
            "                                                                 reshape_64[0][0]                 \n",
            "                                                                 reshape_65[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "lambda_2 (Lambda)               (None, 1152, 8)      0           concatenate_2[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "digitcapsule (CapsuleLayer)     (None, 1152, 8)      1474560     lambda_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "mask_4 (Mask)                   (None, 1152, 8)      0           digitcapsule[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "flatten_4 (Flatten)             (None, 9216)         0           mask_4[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "encoder (Length)                (None, 1152)         0           digitcapsule[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "decoder (Sequential)            (None, 28, 28, 1)    1411344     flatten_4[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 8,215,568\n",
            "Trainable params: 8,215,568\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D9qYRmxtSwKw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#encoder_loss\n",
        "def margin_loss(y_true, y_pred):\n",
        "  m_plus=0.9\n",
        "  m_minus=0.1\n",
        "\n",
        "  L = y_true * K.square(K.maximum(0., m_plus - y_pred)) + \\\n",
        "        0.5 * (1 - y_true) * K.square(K.maximum(0., y_pred - m_minus))\n",
        "  \n",
        "  total=K.sum(L,axis=1)\n",
        "  mean=K.mean(total)\n",
        "\n",
        "  return mean\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i9QP4GuPtlBU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#decoder_loss\n",
        "def MSE(y_true, y_pred):\n",
        "  mse=tf.keras.losses.MeanSquaredError()\n",
        "  return (0.0005)*mse(y_true, y_pred)\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jQP5f_WXSwKd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train.compile(optimizer=\"adam\",loss=[margin_loss,MSE],metrics=[\"accuracy\"])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fznlF06u19C6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau, CSVLogger, LearningRateScheduler"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9z3DOrVF3bqs",
        "colab_type": "code",
        "outputId": "859a2bfb-4ba1-482d-f890-b4ae29ef631b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ltCDDZ3F38mF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import os"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HkbWWgnLWJEC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Include the epoch in the file name (uses `str.format`)\n",
        "\n",
        "checkpoint_path = '/content/drive/My Drive/CapsNet Project/CPnum2.h5'\n",
        "checkpoint_dir = os.path.dirname(checkpoint_path)\n",
        "\n",
        "csv_path ='/content/drive/My Drive/CapsNet Project/CPcsvNum2 - Sheet1.csv'\n",
        "\n",
        "# Create a callback that saves the model's weights every epoch\n",
        "cp_callback = ModelCheckpoint(filepath=checkpoint_path, verbose=1, save_weights_only=True,period=1)\n",
        "csv_callback = CSVLogger(csv_path, separator=\",\", append=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fpHnGXQsjr0u",
        "colab_type": "code",
        "outputId": "801c8915-68cd-4177-9803-e6d5fadffccc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "pip install tensorboardcolab"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: tensorboardcolab in /usr/local/lib/python3.6/dist-packages (0.0.22)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Rb9Xse4Nj773",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorboardcolab import *"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pcA_tWZ3K1I-",
        "colab_type": "code",
        "outputId": "e04b42ea-9708-470e-c2c7-42007e7e5476",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 67
        }
      },
      "source": [
        "tbc=TensorBoardColab()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Wait for 8 seconds...\n",
            "TensorBoard link:\n",
            "https://5b49f116.ngrok.io\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EhfVO03Gqaym",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#es_callback=EarlyStopping(patience=10, restore_best_weights=True)\n",
        "#lr_callback= ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5,min_lr=0.001)\n",
        "def scheduler(epoch,lr):\n",
        "  return float(lr * np.math.exp(-0.1))\n",
        "\n",
        "lr_callback = LearningRateScheduler(scheduler)\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-89pLkWQSwKg",
        "colab_type": "code",
        "outputId": "306d22ce-9a5a-43b5-e2b1-8e1bdb372067",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "train.fit([x_train,y_train],[y_train,x_train],batch_size=128, epochs=50, validation_data=[[x_test, y_test],[y_test, x_test]], callbacks=[cp_callback, csv_callback, lr_callback, TensorBoardColabCallback(tbc)])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/math_grad.py:1424: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.where in 2.0, which has the same broadcast rule as np.where\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:986: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:973: The name tf.assign is deprecated. Please use tf.compat.v1.assign instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:2741: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.\n",
            "\n",
            "Train on 60000 samples, validate on 10000 samples\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorboardcolab/core.py:49: The name tf.summary.FileWriter is deprecated. Please use tf.compat.v1.summary.FileWriter instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:174: The name tf.get_default_session is deprecated. Please use tf.compat.v1.get_default_session instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:181: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:190: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:199: The name tf.is_variable_initialized is deprecated. Please use tf.compat.v1.is_variable_initialized instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:206: The name tf.variables_initializer is deprecated. Please use tf.compat.v1.variables_initializer instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/callbacks.py:850: The name tf.summary.merge_all is deprecated. Please use tf.compat.v1.summary.merge_all instead.\n",
            "\n",
            "Epoch 1/50\n",
            "60000/60000 [==============================] - 269s 4ms/step - loss: 0.2341 - encoder_loss: 0.2340 - decoder_loss: 7.1018e-05 - encoder_acc: 0.7477 - decoder_acc: 0.7870 - val_loss: 0.0245 - val_encoder_loss: 0.0245 - val_decoder_loss: 3.3645e-05 - val_encoder_acc: 0.9841 - val_decoder_acc: 0.8036\n",
            "\n",
            "Epoch 00001: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorboardcolab/callbacks.py:51: The name tf.Summary is deprecated. Please use tf.compat.v1.Summary instead.\n",
            "\n",
            "Epoch 2/50\n",
            "60000/60000 [==============================] - 261s 4ms/step - loss: 0.0222 - encoder_loss: 0.0222 - decoder_loss: 3.3105e-05 - encoder_acc: 0.9855 - decoder_acc: 0.8043 - val_loss: 0.0175 - val_encoder_loss: 0.0175 - val_decoder_loss: 3.2769e-05 - val_encoder_acc: 0.9869 - val_decoder_acc: 0.8030\n",
            "\n",
            "Epoch 00002: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 3/50\n",
            "60000/60000 [==============================] - 261s 4ms/step - loss: 0.0157 - encoder_loss: 0.0157 - decoder_loss: 3.2090e-05 - encoder_acc: 0.9897 - decoder_acc: 0.8052 - val_loss: 0.0153 - val_encoder_loss: 0.0153 - val_decoder_loss: 3.1308e-05 - val_encoder_acc: 0.9903 - val_decoder_acc: 0.8041\n",
            "\n",
            "Epoch 00003: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 4/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 0.0120 - encoder_loss: 0.0120 - decoder_loss: 3.0164e-05 - encoder_acc: 0.9924 - decoder_acc: 0.8036 - val_loss: 0.0122 - val_encoder_loss: 0.0122 - val_decoder_loss: 2.9039e-05 - val_encoder_acc: 0.9911 - val_decoder_acc: 0.7998\n",
            "\n",
            "Epoch 00004: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 5/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 0.0092 - encoder_loss: 0.0091 - decoder_loss: 2.8406e-05 - encoder_acc: 0.9945 - decoder_acc: 0.7990 - val_loss: 0.0102 - val_encoder_loss: 0.0102 - val_decoder_loss: 2.7805e-05 - val_encoder_acc: 0.9922 - val_decoder_acc: 0.7974\n",
            "\n",
            "Epoch 00005: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 6/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 0.0072 - encoder_loss: 0.0071 - decoder_loss: 2.7637e-05 - encoder_acc: 0.9959 - decoder_acc: 0.7971 - val_loss: 0.0096 - val_encoder_loss: 0.0095 - val_decoder_loss: 2.7290e-05 - val_encoder_acc: 0.9932 - val_decoder_acc: 0.7967\n",
            "\n",
            "Epoch 00006: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 7/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 0.0055 - encoder_loss: 0.0055 - decoder_loss: 2.7283e-05 - encoder_acc: 0.9968 - decoder_acc: 0.7961 - val_loss: 0.0088 - val_encoder_loss: 0.0088 - val_decoder_loss: 2.6995e-05 - val_encoder_acc: 0.9928 - val_decoder_acc: 0.7957\n",
            "\n",
            "Epoch 00007: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 8/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 0.0042 - encoder_loss: 0.0042 - decoder_loss: 2.7088e-05 - encoder_acc: 0.9978 - decoder_acc: 0.7954 - val_loss: 0.0091 - val_encoder_loss: 0.0091 - val_decoder_loss: 2.6873e-05 - val_encoder_acc: 0.9927 - val_decoder_acc: 0.7953\n",
            "\n",
            "Epoch 00008: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 9/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 0.0034 - encoder_loss: 0.0033 - decoder_loss: 2.6975e-05 - encoder_acc: 0.9986 - decoder_acc: 0.7950 - val_loss: 0.0078 - val_encoder_loss: 0.0077 - val_decoder_loss: 2.6744e-05 - val_encoder_acc: 0.9928 - val_decoder_acc: 0.7945\n",
            "\n",
            "Epoch 00009: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 10/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 0.0024 - encoder_loss: 0.0024 - decoder_loss: 2.6908e-05 - encoder_acc: 0.9990 - decoder_acc: 0.7946 - val_loss: 0.0074 - val_encoder_loss: 0.0074 - val_decoder_loss: 2.6674e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7938\n",
            "\n",
            "Epoch 00010: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 11/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 0.0018 - encoder_loss: 0.0018 - decoder_loss: 2.6866e-05 - encoder_acc: 0.9994 - decoder_acc: 0.7943 - val_loss: 0.0073 - val_encoder_loss: 0.0073 - val_decoder_loss: 2.6647e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7943\n",
            "\n",
            "Epoch 00011: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 12/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 0.0014 - encoder_loss: 0.0013 - decoder_loss: 2.6840e-05 - encoder_acc: 0.9996 - decoder_acc: 0.7941 - val_loss: 0.0070 - val_encoder_loss: 0.0070 - val_decoder_loss: 2.6628e-05 - val_encoder_acc: 0.9928 - val_decoder_acc: 0.7944\n",
            "\n",
            "Epoch 00012: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 13/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 0.0010 - encoder_loss: 0.0010 - decoder_loss: 2.6821e-05 - encoder_acc: 0.9997 - decoder_acc: 0.7940 - val_loss: 0.0074 - val_encoder_loss: 0.0073 - val_decoder_loss: 2.6576e-05 - val_encoder_acc: 0.9929 - val_decoder_acc: 0.7934\n",
            "\n",
            "Epoch 00013: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 14/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 7.6907e-04 - encoder_loss: 7.4227e-04 - decoder_loss: 2.6807e-05 - encoder_acc: 0.9998 - decoder_acc: 0.7938 - val_loss: 0.0068 - val_encoder_loss: 0.0067 - val_decoder_loss: 2.6581e-05 - val_encoder_acc: 0.9934 - val_decoder_acc: 0.7939\n",
            "\n",
            "Epoch 00014: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 15/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 5.7690e-04 - encoder_loss: 5.5011e-04 - decoder_loss: 2.6795e-05 - encoder_acc: 0.9999 - decoder_acc: 0.7938 - val_loss: 0.0069 - val_encoder_loss: 0.0068 - val_decoder_loss: 2.6565e-05 - val_encoder_acc: 0.9934 - val_decoder_acc: 0.7940\n",
            "\n",
            "Epoch 00015: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 16/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 4.3269e-04 - encoder_loss: 4.0590e-04 - decoder_loss: 2.6786e-05 - encoder_acc: 0.9999 - decoder_acc: 0.7937 - val_loss: 0.0070 - val_encoder_loss: 0.0069 - val_decoder_loss: 2.6562e-05 - val_encoder_acc: 0.9942 - val_decoder_acc: 0.7938\n",
            "\n",
            "Epoch 00016: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 17/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 3.5085e-04 - encoder_loss: 3.2407e-04 - decoder_loss: 2.6778e-05 - encoder_acc: 0.9999 - decoder_acc: 0.7937 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6542e-05 - val_encoder_acc: 0.9932 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00017: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 18/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 2.4184e-04 - encoder_loss: 2.1507e-04 - decoder_loss: 2.6772e-05 - encoder_acc: 0.9999 - decoder_acc: 0.7937 - val_loss: 0.0064 - val_encoder_loss: 0.0064 - val_decoder_loss: 2.6555e-05 - val_encoder_acc: 0.9937 - val_decoder_acc: 0.7941\n",
            "\n",
            "Epoch 00018: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 19/50\n",
            "60000/60000 [==============================] - 259s 4ms/step - loss: 1.9287e-04 - encoder_loss: 1.6610e-04 - decoder_loss: 2.6768e-05 - encoder_acc: 0.9999 - decoder_acc: 0.7937 - val_loss: 0.0066 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6532e-05 - val_encoder_acc: 0.9938 - val_decoder_acc: 0.7937\n",
            "\n",
            "Epoch 00019: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 20/50\n",
            "60000/60000 [==============================] - 261s 4ms/step - loss: 1.5187e-04 - encoder_loss: 1.2510e-04 - decoder_loss: 2.6764e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7937 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6531e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7938\n",
            "\n",
            "Epoch 00020: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 21/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 1.3304e-04 - encoder_loss: 1.0628e-04 - decoder_loss: 2.6760e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7937 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6524e-05 - val_encoder_acc: 0.9934 - val_decoder_acc: 0.7936\n",
            "\n",
            "Epoch 00021: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 22/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 1.0760e-04 - encoder_loss: 8.0841e-05 - decoder_loss: 2.6757e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0063 - val_encoder_loss: 0.0063 - val_decoder_loss: 2.6529e-05 - val_encoder_acc: 0.9939 - val_decoder_acc: 0.7936\n",
            "\n",
            "Epoch 00022: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 23/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 1.0505e-04 - encoder_loss: 7.8291e-05 - decoder_loss: 2.6754e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6521e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7937\n",
            "\n",
            "Epoch 00023: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 24/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 7.3742e-05 - encoder_loss: 4.6990e-05 - decoder_loss: 2.6751e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7937 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6528e-05 - val_encoder_acc: 0.9938 - val_decoder_acc: 0.7938\n",
            "\n",
            "Epoch 00024: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 25/50\n",
            "60000/60000 [==============================] - 259s 4ms/step - loss: 5.6867e-05 - encoder_loss: 3.0117e-05 - decoder_loss: 2.6750e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7937 - val_loss: 0.0064 - val_encoder_loss: 0.0064 - val_decoder_loss: 2.6515e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00025: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 26/50\n",
            "60000/60000 [==============================] - 260s 4ms/step - loss: 4.6775e-05 - encoder_loss: 2.0027e-05 - decoder_loss: 2.6749e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0064 - val_encoder_loss: 0.0064 - val_decoder_loss: 2.6513e-05 - val_encoder_acc: 0.9941 - val_decoder_acc: 0.7934\n",
            "\n",
            "Epoch 00026: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 27/50\n",
            "60000/60000 [==============================] - 259s 4ms/step - loss: 4.3366e-05 - encoder_loss: 1.6619e-05 - decoder_loss: 2.6747e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0064 - val_encoder_loss: 0.0063 - val_decoder_loss: 2.6513e-05 - val_encoder_acc: 0.9938 - val_decoder_acc: 0.7936\n",
            "\n",
            "Epoch 00027: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 28/50\n",
            "60000/60000 [==============================] - 260s 4ms/step - loss: 3.9263e-05 - encoder_loss: 1.2517e-05 - decoder_loss: 2.6746e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6520e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7939\n",
            "\n",
            "Epoch 00028: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 29/50\n",
            "60000/60000 [==============================] - 260s 4ms/step - loss: 3.6049e-05 - encoder_loss: 9.3045e-06 - decoder_loss: 2.6745e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0064 - val_encoder_loss: 0.0064 - val_decoder_loss: 2.6516e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7937\n",
            "\n",
            "Epoch 00029: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 30/50\n",
            "60000/60000 [==============================] - 260s 4ms/step - loss: 3.3906e-05 - encoder_loss: 7.1630e-06 - decoder_loss: 2.6743e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7937 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6506e-05 - val_encoder_acc: 0.9939 - val_decoder_acc: 0.7934\n",
            "\n",
            "Epoch 00030: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 31/50\n",
            "60000/60000 [==============================] - 260s 4ms/step - loss: 3.2314e-05 - encoder_loss: 5.5714e-06 - decoder_loss: 2.6743e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0064 - val_decoder_loss: 2.6509e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00031: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 32/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 3.2587e-05 - encoder_loss: 5.8459e-06 - decoder_loss: 2.6742e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0064 - val_decoder_loss: 2.6508e-05 - val_encoder_acc: 0.9938 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00032: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 33/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 3.1434e-05 - encoder_loss: 4.6935e-06 - decoder_loss: 2.6740e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6506e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00033: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 34/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 3.0009e-05 - encoder_loss: 3.2694e-06 - decoder_loss: 2.6740e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6506e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7934\n",
            "\n",
            "Epoch 00034: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 35/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 2.9854e-05 - encoder_loss: 3.1150e-06 - decoder_loss: 2.6739e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6507e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00035: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 36/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 2.9100e-05 - encoder_loss: 2.3611e-06 - decoder_loss: 2.6739e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6507e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7934\n",
            "\n",
            "Epoch 00036: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 37/50\n",
            "60000/60000 [==============================] - 255s 4ms/step - loss: 2.8416e-05 - encoder_loss: 1.6781e-06 - decoder_loss: 2.6738e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0065 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6506e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00037: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 38/50\n",
            "60000/60000 [==============================] - 255s 4ms/step - loss: 2.8158e-05 - encoder_loss: 1.4200e-06 - decoder_loss: 2.6738e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0065 - val_decoder_loss: 2.6503e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00038: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 39/50\n",
            "60000/60000 [==============================] - 259s 4ms/step - loss: 2.7930e-05 - encoder_loss: 1.1928e-06 - decoder_loss: 2.6737e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6503e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00039: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 40/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 2.7777e-05 - encoder_loss: 1.0403e-06 - decoder_loss: 2.6737e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6506e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00040: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 41/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 2.7607e-05 - encoder_loss: 8.7049e-07 - decoder_loss: 2.6737e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6504e-05 - val_encoder_acc: 0.9934 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00041: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 42/50\n",
            "60000/60000 [==============================] - 255s 4ms/step - loss: 2.7464e-05 - encoder_loss: 7.2738e-07 - decoder_loss: 2.6736e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6504e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00042: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 43/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 2.7439e-05 - encoder_loss: 7.0325e-07 - decoder_loss: 2.6736e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0066 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6504e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00043: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 44/50\n",
            "60000/60000 [==============================] - 255s 4ms/step - loss: 2.7288e-05 - encoder_loss: 5.5233e-07 - decoder_loss: 2.6736e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6505e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00044: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 45/50\n",
            "60000/60000 [==============================] - 255s 4ms/step - loss: 2.7263e-05 - encoder_loss: 5.2777e-07 - decoder_loss: 2.6735e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6504e-05 - val_encoder_acc: 0.9938 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00045: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 46/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 2.7232e-05 - encoder_loss: 4.9699e-07 - decoder_loss: 2.6735e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0066 - val_decoder_loss: 2.6504e-05 - val_encoder_acc: 0.9934 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00046: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 47/50\n",
            "60000/60000 [==============================] - 257s 4ms/step - loss: 2.7141e-05 - encoder_loss: 4.0604e-07 - decoder_loss: 2.6735e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0067 - val_decoder_loss: 2.6504e-05 - val_encoder_acc: 0.9936 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00047: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 48/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 2.7100e-05 - encoder_loss: 3.6500e-07 - decoder_loss: 2.6735e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0067 - val_decoder_loss: 2.6503e-05 - val_encoder_acc: 0.9933 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00048: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 49/50\n",
            "60000/60000 [==============================] - 256s 4ms/step - loss: 2.7068e-05 - encoder_loss: 3.3378e-07 - decoder_loss: 2.6735e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0067 - val_decoder_loss: 2.6503e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00049: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n",
            "Epoch 50/50\n",
            "60000/60000 [==============================] - 258s 4ms/step - loss: 2.7046e-05 - encoder_loss: 3.1209e-07 - decoder_loss: 2.6734e-05 - encoder_acc: 1.0000 - decoder_acc: 0.7936 - val_loss: 0.0067 - val_encoder_loss: 0.0067 - val_decoder_loss: 2.6503e-05 - val_encoder_acc: 0.9935 - val_decoder_acc: 0.7935\n",
            "\n",
            "Epoch 00050: saving model to /content/drive/My Drive/CapsNet Project/CPnum2.h5\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fe2a2061550>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 66
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3O0fpYCeSwKi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        " y_pred, x_recon = evalm.predict(x_test, batch_size=100)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tvMxqdgrXbx1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "############ RESULTS"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vINWrc0kX_jL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pred=np.zeros((10000,1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yMQLLgUsStN8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in range(len(y_pred)):\n",
        "  pred[i]=np.argmax(y_pred[i])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7kSfre1ASwKx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.metrics import confusion_matrix, classification_report"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "82rKym1VXloU",
        "colab_type": "code",
        "outputId": "7e992927-ab31-4c47-b64a-1293750c249e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        }
      },
      "source": [
        "#test_classification_accuracy\n",
        "print(classification_report(np.argmax(y_test, axis = 1), pred))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.99      1.00      1.00       980\n",
            "           1       0.99      1.00      1.00      1135\n",
            "           2       1.00      0.99      0.99      1032\n",
            "           3       0.99      1.00      0.99      1010\n",
            "           4       0.99      0.99      0.99       982\n",
            "           5       1.00      0.99      0.99       892\n",
            "           6       0.99      0.99      0.99       958\n",
            "           7       0.99      0.99      0.99      1028\n",
            "           8       1.00      0.99      0.99       974\n",
            "           9       0.99      0.99      0.99      1009\n",
            "\n",
            "    accuracy                           0.99     10000\n",
            "   macro avg       0.99      0.99      0.99     10000\n",
            "weighted avg       0.99      0.99      0.99     10000\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FXSab8GFnGbF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from random import randint "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a9Ws_fZVmUVh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "random=[]\n",
        "for _ in range(9):\n",
        "\tvalue = randint(0, len(x_test))\n",
        "\trandom.append(value)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RbQb72gYXrfV",
        "colab_type": "code",
        "outputId": "d3a1f9f2-ffaa-4458-c2cd-c03ca04d41ba",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 553
        }
      },
      "source": [
        "#test_image_reconstuction\n",
        "from matplotlib import pyplot\n",
        "# load dataset\n",
        "\n",
        "# summarize loaded dataset\n",
        "# plot first few images\n",
        "print(\"original\")\n",
        "for i in range(9):\n",
        "\t# define subplot\n",
        "\tpyplot.subplot(330 + 1 + i)\n",
        "\tpyplot.imshow(Xt_test_img[random[i]])\n",
        "# show the figure\n",
        "pyplot.show()\n",
        "print(\"reconstructed\")\n",
        "for i in range(9):\n",
        "\t# define subplot\n",
        "\tpyplot.subplot(330 + 1 + i)\n",
        "\t# plot raw pixel data\n",
        " \n",
        "\tpyplot.imshow(x_recon[random[i]].reshape(28,28))\n",
        "# show the figure\n",
        "pyplot.show()\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "original\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "reconstructed\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yy65ADOnUEYO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}