agrhn · January 21, 2021 00:03 · ReefSami9 · Apr 3, 2021
diff --git a/googlenet_tensorflow.ipynb b/googlenet_tensorflow.ipynb
 {
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "GoogLeNet_Tensorflow.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyP/t+xJaeMRXGPvggde5FaD",
      "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/gist/mrgrhn/a6ad98dbc81f3ec8f73d39415452de9a/googlenet_tensorflow.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CXt05UtlLUcz"
      },
      "source": [
        "import tensorflow as tf\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "from tensorflow.keras import datasets, layers, models, losses, Model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "baxvTVSQLcC9"
      },
      "source": [
        "(x_train, y_train), (x_test, y_test)=tf.keras.datasets.mnist.load_data()\r\n",
        "x_train = tf.pad(x_train, [[0, 0], [2,2], [2,2]])/255\r\n",
        "x_test = tf.pad(x_test, [[0, 0], [2,2], [2,2]])/255\r\n",
        "x_train = tf.expand_dims(x_train, axis=3, name=None)\r\n",
        "x_test = tf.expand_dims(x_test, axis=3, name=None)\r\n",
        "x_train = tf.repeat(x_train, 3, axis=3)\r\n",
        "x_test = tf.repeat(x_test, 3, axis=3)\r\n",
        "x_val = x_train[-2000:,:,:]\r\n",
        "y_val = y_train[-2000:]\r\n",
        "x_train = x_train[:-2000,:,:]\r\n",
        "y_train = y_train[:-2000]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6wKcYCdaLcFU"
      },
      "source": [
        "def inception(x,\r\n",
        "              filters_1x1,\r\n",
        "              filters_3x3_reduce,\r\n",
        "              filters_3x3,\r\n",
        "              filters_5x5_reduce,\r\n",
        "              filters_5x5,\r\n",
        "              filters_pool):\r\n",
        "  path1 = layers.Conv2D(filters_1x1, (1, 1), padding='same', activation='relu')(x)\r\n",
        "\r\n",
        "  path2 = layers.Conv2D(filters_3x3_reduce, (1, 1), padding='same', activation='relu')(x)\r\n",
        "  path2 = layers.Conv2D(filters_3x3, (1, 1), padding='same', activation='relu')(path2)\r\n",
        "\r\n",
        "  path3 = layers.Conv2D(filters_5x5_reduce, (1, 1), padding='same', activation='relu')(x)\r\n",
        "  path3 = layers.Conv2D(filters_5x5, (1, 1), padding='same', activation='relu')(path3)\r\n",
        "\r\n",
        "  path4 = layers.MaxPool2D((3, 3), strides=(1, 1), padding='same')(x)\r\n",
        "  path4 = layers.Conv2D(filters_pool, (1, 1), padding='same', activation='relu')(path4)\r\n",
        "\r\n",
        "  return tf.concat([path1, path2, path3, path4], axis=3)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nDEI1TMtLcHq"
      },
      "source": [
        "inp = layers.Input(shape=(32, 32, 3))\r\n",
        "input_tensor = layers.experimental.preprocessing.Resizing(224, 224, interpolation=\"bilinear\", input_shape=x_train.shape[1:])(inp)\r\n",
        "\r\n",
        "x = layers.Conv2D(64, 7, strides=2, padding='same', activation='relu')(input_tensor)\r\n",
        "x = layers.MaxPooling2D(3, strides=2)(x)\r\n",
        "\r\n",
        "x = layers.Conv2D(64, 1, strides=1, padding='same', activation='relu')(x)\r\n",
        "x = layers.Conv2D(192, 3, strides=1, padding='same', activation='relu')(x)\r\n",
        "\r\n",
        "x = layers.MaxPooling2D(3, strides=2)(x)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=64,\r\n",
        "              filters_3x3_reduce=96,\r\n",
        "              filters_3x3=128,\r\n",
        "              filters_5x5_reduce=16,\r\n",
        "              filters_5x5=32,\r\n",
        "              filters_pool=32)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=128,\r\n",
        "              filters_3x3_reduce=128,\r\n",
        "              filters_3x3=192,\r\n",
        "              filters_5x5_reduce=32,\r\n",
        "              filters_5x5=96,\r\n",
        "              filters_pool=64)\r\n",
        "\r\n",
        "x = layers.MaxPooling2D(3, strides=2)(x)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=192,\r\n",
        "              filters_3x3_reduce=96,\r\n",
        "              filters_3x3=208,\r\n",
        "              filters_5x5_reduce=16,\r\n",
        "              filters_5x5=48,\r\n",
        "              filters_pool=64)\r\n",
        "\r\n",
        "aux1 = layers.AveragePooling2D((5, 5), strides=3)(x)\r\n",
        "aux1 = layers.Conv2D(128, 1, padding='same', activation='relu')(aux1)\r\n",
        "aux1 = layers.Flatten()(aux1)\r\n",
        "aux1 = layers.Dense(1024, activation='relu')(aux1)\r\n",
        "aux1 = layers.Dropout(0.7)(aux1)\r\n",
        "aux1 = layers.Dense(10, activation='softmax')(aux1)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=160,\r\n",
        "              filters_3x3_reduce=112,\r\n",
        "              filters_3x3=224,\r\n",
        "              filters_5x5_reduce=24,\r\n",
        "              filters_5x5=64,\r\n",
        "              filters_pool=64)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=128,\r\n",
        "              filters_3x3_reduce=128,\r\n",
        "              filters_3x3=256,\r\n",
        "              filters_5x5_reduce=24,\r\n",
        "              filters_5x5=64,\r\n",
        "              filters_pool=64)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=112,\r\n",
        "              filters_3x3_reduce=144,\r\n",
        "              filters_3x3=288,\r\n",
        "              filters_5x5_reduce=32,\r\n",
        "              filters_5x5=64,\r\n",
        "              filters_pool=64)\r\n",
        "\r\n",
        "aux2 = layers.AveragePooling2D((5, 5), strides=3)(x)\r\n",
        "aux2 = layers.Conv2D(128, 1, padding='same', activation='relu')(aux2)\r\n",
        "aux2 = layers.Flatten()(aux2)\r\n",
        "aux2 = layers.Dense(1024, activation='relu')(aux2)\r\n",
        "aux2 = layers.Dropout(0.7)(aux2)\r\n",
        "aux2 = layers.Dense(10, activation='softmax')(aux2)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=256,\r\n",
        "              filters_3x3_reduce=160,\r\n",
        "              filters_3x3=320,\r\n",
        "              filters_5x5_reduce=32,\r\n",
        "              filters_5x5=128,\r\n",
        "              filters_pool=128)\r\n",
        "\r\n",
        "x = layers.MaxPooling2D(3, strides=2)(x)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=256,\r\n",
        "              filters_3x3_reduce=160,\r\n",
        "              filters_3x3=320,\r\n",
        "              filters_5x5_reduce=32,\r\n",
        "              filters_5x5=128,\r\n",
        "              filters_pool=128)\r\n",
        "\r\n",
        "x = inception(x,\r\n",
        "              filters_1x1=384,\r\n",
        "              filters_3x3_reduce=192,\r\n",
        "              filters_3x3=384,\r\n",
        "              filters_5x5_reduce=48,\r\n",
        "              filters_5x5=128,\r\n",
        "              filters_pool=128)\r\n",
        "\r\n",
        "x = layers.GlobalAveragePooling2D()(x)\r\n",
        "\r\n",
        "x = layers.Dropout(0.4)(x)\r\n",
        "out = layers.Dense(10, activation='softmax')(x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WzIFlLP9LcLQ"
      },
      "source": [
        "model = Model(inputs = inp, outputs = [out, aux1, aux2])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wJjdIxC1aWwO"
      },
      "source": [
        "model.compile(optimizer='adam', loss=[losses.sparse_categorical_crossentropy, losses.sparse_categorical_crossentropy, losses.sparse_categorical_crossentropy], loss_weights=[1, 0.3, 0.3], metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VofnKKKoauYX",
        "outputId": "4d72e8d4-219e-4908-ec1a-c609d78d3b3c"
      },
      "source": [
        "history = model.fit(x_train, [y_train, y_train, y_train], validation_data=(x_val, [y_val, y_val, y_val]), batch_size=64, epochs=40)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/40\n",
            "907/907 [==============================] - 159s 168ms/step - loss: 2.1553 - dense_4_loss: 1.5218 - dense_1_loss: 1.0258 - dense_3_loss: 1.0859 - dense_4_accuracy: 0.4329 - dense_1_accuracy: 0.6360 - dense_3_accuracy: 0.6135 - val_loss: 0.1934 - val_dense_4_loss: 0.1529 - val_dense_1_loss: 0.0614 - val_dense_3_loss: 0.0736 - val_dense_4_accuracy: 0.9530 - val_dense_1_accuracy: 0.9805 - val_dense_3_accuracy: 0.9815\n",
            "Epoch 2/40\n",
            "907/907 [==============================] - 151s 166ms/step - loss: 0.2582 - dense_4_loss: 0.1851 - dense_1_loss: 0.1179 - dense_3_loss: 0.1260 - dense_4_accuracy: 0.9444 - dense_1_accuracy: 0.9640 - dense_3_accuracy: 0.9620 - val_loss: 0.1014 - val_dense_4_loss: 0.0771 - val_dense_1_loss: 0.0410 - val_dense_3_loss: 0.0402 - val_dense_4_accuracy: 0.9780 - val_dense_1_accuracy: 0.9890 - val_dense_3_accuracy: 0.9895\n",
            "Epoch 3/40\n",
            "907/907 [==============================] - 150s 165ms/step - loss: 0.1458 - dense_4_loss: 0.0968 - dense_1_loss: 0.0798 - dense_3_loss: 0.0834 - dense_4_accuracy: 0.9710 - dense_1_accuracy: 0.9762 - dense_3_accuracy: 0.9741 - val_loss: 0.0826 - val_dense_4_loss: 0.0557 - val_dense_1_loss: 0.0445 - val_dense_3_loss: 0.0451 - val_dense_4_accuracy: 0.9870 - val_dense_1_accuracy: 0.9870 - val_dense_3_accuracy: 0.9880\n",
            "Epoch 4/40\n",
            "907/907 [==============================] - 150s 165ms/step - loss: 0.1121 - dense_4_loss: 0.0750 - dense_1_loss: 0.0606 - dense_3_loss: 0.0629 - dense_4_accuracy: 0.9770 - dense_1_accuracy: 0.9809 - dense_3_accuracy: 0.9798 - val_loss: 0.0652 - val_dense_4_loss: 0.0475 - val_dense_1_loss: 0.0287 - val_dense_3_loss: 0.0302 - val_dense_4_accuracy: 0.9890 - val_dense_1_accuracy: 0.9930 - val_dense_3_accuracy: 0.9950\n",
            "Epoch 5/40\n",
            "907/907 [==============================] - 149s 165ms/step - loss: 0.0928 - dense_4_loss: 0.0610 - dense_1_loss: 0.0500 - dense_3_loss: 0.0560 - dense_4_accuracy: 0.9811 - dense_1_accuracy: 0.9855 - dense_3_accuracy: 0.9839 - val_loss: 0.0695 - val_dense_4_loss: 0.0499 - val_dense_1_loss: 0.0335 - val_dense_3_loss: 0.0320 - val_dense_4_accuracy: 0.9860 - val_dense_1_accuracy: 0.9930 - val_dense_3_accuracy: 0.9930\n",
            "Epoch 6/40\n",
            "907/907 [==============================] - 149s 164ms/step - loss: 0.0856 - dense_4_loss: 0.0569 - dense_1_loss: 0.0467 - dense_3_loss: 0.0490 - dense_4_accuracy: 0.9815 - dense_1_accuracy: 0.9851 - dense_3_accuracy: 0.9852 - val_loss: 0.0921 - val_dense_4_loss: 0.0639 - val_dense_1_loss: 0.0515 - val_dense_3_loss: 0.0422 - val_dense_4_accuracy: 0.9840 - val_dense_1_accuracy: 0.9895 - val_dense_3_accuracy: 0.9885\n",
            "Epoch 7/40\n",
            "907/907 [==============================] - 149s 164ms/step - loss: 0.0806 - dense_4_loss: 0.0522 - dense_1_loss: 0.0481 - dense_3_loss: 0.0466 - dense_4_accuracy: 0.9841 - dense_1_accuracy: 0.9857 - dense_3_accuracy: 0.9855 - val_loss: 0.0533 - val_dense_4_loss: 0.0390 - val_dense_1_loss: 0.0236 - val_dense_3_loss: 0.0239 - val_dense_4_accuracy: 0.9925 - val_dense_1_accuracy: 0.9945 - val_dense_3_accuracy: 0.9950\n",
            "Epoch 8/40\n",
            "907/907 [==============================] - 149s 164ms/step - loss: 0.0698 - dense_4_loss: 0.0441 - dense_1_loss: 0.0423 - dense_3_loss: 0.0432 - dense_4_accuracy: 0.9866 - dense_1_accuracy: 0.9869 - dense_3_accuracy: 0.9874 - val_loss: 0.0527 - val_dense_4_loss: 0.0357 - val_dense_1_loss: 0.0294 - val_dense_3_loss: 0.0273 - val_dense_4_accuracy: 0.9910 - val_dense_1_accuracy: 0.9950 - val_dense_3_accuracy: 0.9950\n",
            "Epoch 9/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0619 - dense_4_loss: 0.0402 - dense_1_loss: 0.0335 - dense_3_loss: 0.0389 - dense_4_accuracy: 0.9872 - dense_1_accuracy: 0.9893 - dense_3_accuracy: 0.9883 - val_loss: 0.0484 - val_dense_4_loss: 0.0315 - val_dense_1_loss: 0.0309 - val_dense_3_loss: 0.0254 - val_dense_4_accuracy: 0.9930 - val_dense_1_accuracy: 0.9950 - val_dense_3_accuracy: 0.9950\n",
            "Epoch 10/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0602 - dense_4_loss: 0.0400 - dense_1_loss: 0.0326 - dense_3_loss: 0.0348 - dense_4_accuracy: 0.9878 - dense_1_accuracy: 0.9903 - dense_3_accuracy: 0.9896 - val_loss: 0.0398 - val_dense_4_loss: 0.0262 - val_dense_1_loss: 0.0212 - val_dense_3_loss: 0.0242 - val_dense_4_accuracy: 0.9950 - val_dense_1_accuracy: 0.9955 - val_dense_3_accuracy: 0.9960\n",
            "Epoch 11/40\n",
            "907/907 [==============================] - 149s 164ms/step - loss: 0.0509 - dense_4_loss: 0.0316 - dense_1_loss: 0.0314 - dense_3_loss: 0.0328 - dense_4_accuracy: 0.9900 - dense_1_accuracy: 0.9901 - dense_3_accuracy: 0.9899 - val_loss: 0.0832 - val_dense_4_loss: 0.0605 - val_dense_1_loss: 0.0372 - val_dense_3_loss: 0.0385 - val_dense_4_accuracy: 0.9835 - val_dense_1_accuracy: 0.9925 - val_dense_3_accuracy: 0.9915\n",
            "Epoch 12/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0532 - dense_4_loss: 0.0345 - dense_1_loss: 0.0308 - dense_3_loss: 0.0316 - dense_4_accuracy: 0.9893 - dense_1_accuracy: 0.9907 - dense_3_accuracy: 0.9901 - val_loss: 0.0630 - val_dense_4_loss: 0.0475 - val_dense_1_loss: 0.0267 - val_dense_3_loss: 0.0249 - val_dense_4_accuracy: 0.9880 - val_dense_1_accuracy: 0.9950 - val_dense_3_accuracy: 0.9935\n",
            "Epoch 13/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0460 - dense_4_loss: 0.0289 - dense_1_loss: 0.0286 - dense_3_loss: 0.0286 - dense_4_accuracy: 0.9906 - dense_1_accuracy: 0.9916 - dense_3_accuracy: 0.9911 - val_loss: 0.0375 - val_dense_4_loss: 0.0238 - val_dense_1_loss: 0.0194 - val_dense_3_loss: 0.0263 - val_dense_4_accuracy: 0.9955 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9940\n",
            "Epoch 14/40\n",
            "907/907 [==============================] - 149s 164ms/step - loss: 0.0437 - dense_4_loss: 0.0275 - dense_1_loss: 0.0260 - dense_3_loss: 0.0280 - dense_4_accuracy: 0.9915 - dense_1_accuracy: 0.9919 - dense_3_accuracy: 0.9914 - val_loss: 0.0397 - val_dense_4_loss: 0.0254 - val_dense_1_loss: 0.0241 - val_dense_3_loss: 0.0235 - val_dense_4_accuracy: 0.9930 - val_dense_1_accuracy: 0.9945 - val_dense_3_accuracy: 0.9945\n",
            "Epoch 15/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0378 - dense_4_loss: 0.0235 - dense_1_loss: 0.0236 - dense_3_loss: 0.0242 - dense_4_accuracy: 0.9924 - dense_1_accuracy: 0.9925 - dense_3_accuracy: 0.9929 - val_loss: 0.0431 - val_dense_4_loss: 0.0294 - val_dense_1_loss: 0.0247 - val_dense_3_loss: 0.0208 - val_dense_4_accuracy: 0.9945 - val_dense_1_accuracy: 0.9970 - val_dense_3_accuracy: 0.9955\n",
            "Epoch 16/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0351 - dense_4_loss: 0.0227 - dense_1_loss: 0.0212 - dense_3_loss: 0.0202 - dense_4_accuracy: 0.9925 - dense_1_accuracy: 0.9932 - dense_3_accuracy: 0.9939 - val_loss: 0.0412 - val_dense_4_loss: 0.0261 - val_dense_1_loss: 0.0292 - val_dense_3_loss: 0.0212 - val_dense_4_accuracy: 0.9925 - val_dense_1_accuracy: 0.9940 - val_dense_3_accuracy: 0.9965\n",
            "Epoch 17/40\n",
            "907/907 [==============================] - 148s 164ms/step - loss: 0.0331 - dense_4_loss: 0.0200 - dense_1_loss: 0.0215 - dense_3_loss: 0.0222 - dense_4_accuracy: 0.9939 - dense_1_accuracy: 0.9936 - dense_3_accuracy: 0.9931 - val_loss: 0.0534 - val_dense_4_loss: 0.0367 - val_dense_1_loss: 0.0270 - val_dense_3_loss: 0.0287 - val_dense_4_accuracy: 0.9915 - val_dense_1_accuracy: 0.9950 - val_dense_3_accuracy: 0.9945\n",
            "Epoch 18/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0329 - dense_4_loss: 0.0215 - dense_1_loss: 0.0187 - dense_3_loss: 0.0192 - dense_4_accuracy: 0.9931 - dense_1_accuracy: 0.9939 - dense_3_accuracy: 0.9942 - val_loss: 0.0335 - val_dense_4_loss: 0.0214 - val_dense_1_loss: 0.0224 - val_dense_3_loss: 0.0181 - val_dense_4_accuracy: 0.9960 - val_dense_1_accuracy: 0.9970 - val_dense_3_accuracy: 0.9975\n",
            "Epoch 19/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0292 - dense_4_loss: 0.0178 - dense_1_loss: 0.0190 - dense_3_loss: 0.0193 - dense_4_accuracy: 0.9939 - dense_1_accuracy: 0.9945 - dense_3_accuracy: 0.9941 - val_loss: 0.0683 - val_dense_4_loss: 0.0428 - val_dense_1_loss: 0.0383 - val_dense_3_loss: 0.0467 - val_dense_4_accuracy: 0.9880 - val_dense_1_accuracy: 0.9905 - val_dense_3_accuracy: 0.9870\n",
            "Epoch 20/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0319 - dense_4_loss: 0.0199 - dense_1_loss: 0.0201 - dense_3_loss: 0.0198 - dense_4_accuracy: 0.9938 - dense_1_accuracy: 0.9943 - dense_3_accuracy: 0.9943 - val_loss: 0.0395 - val_dense_4_loss: 0.0244 - val_dense_1_loss: 0.0240 - val_dense_3_loss: 0.0262 - val_dense_4_accuracy: 0.9950 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9965\n",
            "Epoch 21/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0291 - dense_4_loss: 0.0178 - dense_1_loss: 0.0189 - dense_3_loss: 0.0188 - dense_4_accuracy: 0.9942 - dense_1_accuracy: 0.9942 - dense_3_accuracy: 0.9942 - val_loss: 0.0403 - val_dense_4_loss: 0.0264 - val_dense_1_loss: 0.0211 - val_dense_3_loss: 0.0250 - val_dense_4_accuracy: 0.9955 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9945\n",
            "Epoch 22/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0256 - dense_4_loss: 0.0156 - dense_1_loss: 0.0150 - dense_3_loss: 0.0183 - dense_4_accuracy: 0.9949 - dense_1_accuracy: 0.9955 - dense_3_accuracy: 0.9947 - val_loss: 0.0393 - val_dense_4_loss: 0.0266 - val_dense_1_loss: 0.0211 - val_dense_3_loss: 0.0214 - val_dense_4_accuracy: 0.9945 - val_dense_1_accuracy: 0.9975 - val_dense_3_accuracy: 0.9960\n",
            "Epoch 23/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0289 - dense_4_loss: 0.0179 - dense_1_loss: 0.0181 - dense_3_loss: 0.0185 - dense_4_accuracy: 0.9941 - dense_1_accuracy: 0.9943 - dense_3_accuracy: 0.9940 - val_loss: 0.0513 - val_dense_4_loss: 0.0340 - val_dense_1_loss: 0.0191 - val_dense_3_loss: 0.0385 - val_dense_4_accuracy: 0.9925 - val_dense_1_accuracy: 0.9975 - val_dense_3_accuracy: 0.9955\n",
            "Epoch 24/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0276 - dense_4_loss: 0.0177 - dense_1_loss: 0.0165 - dense_3_loss: 0.0165 - dense_4_accuracy: 0.9948 - dense_1_accuracy: 0.9949 - dense_3_accuracy: 0.9948 - val_loss: 0.0404 - val_dense_4_loss: 0.0259 - val_dense_1_loss: 0.0220 - val_dense_3_loss: 0.0263 - val_dense_4_accuracy: 0.9940 - val_dense_1_accuracy: 0.9955 - val_dense_3_accuracy: 0.9960\n",
            "Epoch 25/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0250 - dense_4_loss: 0.0149 - dense_1_loss: 0.0159 - dense_3_loss: 0.0178 - dense_4_accuracy: 0.9950 - dense_1_accuracy: 0.9954 - dense_3_accuracy: 0.9946 - val_loss: 0.0493 - val_dense_4_loss: 0.0342 - val_dense_1_loss: 0.0228 - val_dense_3_loss: 0.0277 - val_dense_4_accuracy: 0.9930 - val_dense_1_accuracy: 0.9950 - val_dense_3_accuracy: 0.9935\n",
            "Epoch 26/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0258 - dense_4_loss: 0.0162 - dense_1_loss: 0.0143 - dense_3_loss: 0.0179 - dense_4_accuracy: 0.9947 - dense_1_accuracy: 0.9955 - dense_3_accuracy: 0.9948 - val_loss: 0.0599 - val_dense_4_loss: 0.0431 - val_dense_1_loss: 0.0280 - val_dense_3_loss: 0.0280 - val_dense_4_accuracy: 0.9895 - val_dense_1_accuracy: 0.9970 - val_dense_3_accuracy: 0.9915\n",
            "Epoch 27/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0228 - dense_4_loss: 0.0147 - dense_1_loss: 0.0147 - dense_3_loss: 0.0123 - dense_4_accuracy: 0.9954 - dense_1_accuracy: 0.9954 - dense_3_accuracy: 0.9961 - val_loss: 0.0342 - val_dense_4_loss: 0.0176 - val_dense_1_loss: 0.0248 - val_dense_3_loss: 0.0304 - val_dense_4_accuracy: 0.9960 - val_dense_1_accuracy: 0.9975 - val_dense_3_accuracy: 0.9950\n",
            "Epoch 28/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0227 - dense_4_loss: 0.0135 - dense_1_loss: 0.0166 - dense_3_loss: 0.0140 - dense_4_accuracy: 0.9953 - dense_1_accuracy: 0.9949 - dense_3_accuracy: 0.9959 - val_loss: 0.0485 - val_dense_4_loss: 0.0332 - val_dense_1_loss: 0.0255 - val_dense_3_loss: 0.0254 - val_dense_4_accuracy: 0.9940 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9960\n",
            "Epoch 29/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0263 - dense_4_loss: 0.0165 - dense_1_loss: 0.0161 - dense_3_loss: 0.0168 - dense_4_accuracy: 0.9943 - dense_1_accuracy: 0.9948 - dense_3_accuracy: 0.9951 - val_loss: 0.0538 - val_dense_4_loss: 0.0318 - val_dense_1_loss: 0.0270 - val_dense_3_loss: 0.0466 - val_dense_4_accuracy: 0.9945 - val_dense_1_accuracy: 0.9955 - val_dense_3_accuracy: 0.9910\n",
            "Epoch 30/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0271 - dense_4_loss: 0.0167 - dense_1_loss: 0.0171 - dense_3_loss: 0.0177 - dense_4_accuracy: 0.9943 - dense_1_accuracy: 0.9950 - dense_3_accuracy: 0.9944 - val_loss: 0.0480 - val_dense_4_loss: 0.0284 - val_dense_1_loss: 0.0314 - val_dense_3_loss: 0.0340 - val_dense_4_accuracy: 0.9970 - val_dense_1_accuracy: 0.9955 - val_dense_3_accuracy: 0.9955\n",
            "Epoch 31/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0169 - dense_4_loss: 0.0098 - dense_1_loss: 0.0117 - dense_3_loss: 0.0120 - dense_4_accuracy: 0.9967 - dense_1_accuracy: 0.9966 - dense_3_accuracy: 0.9961 - val_loss: 0.0511 - val_dense_4_loss: 0.0309 - val_dense_1_loss: 0.0347 - val_dense_3_loss: 0.0329 - val_dense_4_accuracy: 0.9955 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9960\n",
            "Epoch 32/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0227 - dense_4_loss: 0.0129 - dense_1_loss: 0.0153 - dense_3_loss: 0.0174 - dense_4_accuracy: 0.9961 - dense_1_accuracy: 0.9956 - dense_3_accuracy: 0.9950 - val_loss: 0.0537 - val_dense_4_loss: 0.0346 - val_dense_1_loss: 0.0283 - val_dense_3_loss: 0.0354 - val_dense_4_accuracy: 0.9930 - val_dense_1_accuracy: 0.9935 - val_dense_3_accuracy: 0.9945\n",
            "Epoch 33/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0230 - dense_4_loss: 0.0144 - dense_1_loss: 0.0145 - dense_3_loss: 0.0142 - dense_4_accuracy: 0.9956 - dense_1_accuracy: 0.9963 - dense_3_accuracy: 0.9953 - val_loss: 0.0355 - val_dense_4_loss: 0.0223 - val_dense_1_loss: 0.0182 - val_dense_3_loss: 0.0257 - val_dense_4_accuracy: 0.9960 - val_dense_1_accuracy: 0.9955 - val_dense_3_accuracy: 0.9950\n",
            "Epoch 34/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0167 - dense_4_loss: 0.0095 - dense_1_loss: 0.0123 - dense_3_loss: 0.0115 - dense_4_accuracy: 0.9967 - dense_1_accuracy: 0.9964 - dense_3_accuracy: 0.9962 - val_loss: 0.0431 - val_dense_4_loss: 0.0275 - val_dense_1_loss: 0.0238 - val_dense_3_loss: 0.0282 - val_dense_4_accuracy: 0.9955 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9955\n",
            "Epoch 35/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0186 - dense_4_loss: 0.0105 - dense_1_loss: 0.0144 - dense_3_loss: 0.0127 - dense_4_accuracy: 0.9963 - dense_1_accuracy: 0.9955 - dense_3_accuracy: 0.9966 - val_loss: 0.0600 - val_dense_4_loss: 0.0412 - val_dense_1_loss: 0.0265 - val_dense_3_loss: 0.0359 - val_dense_4_accuracy: 0.9935 - val_dense_1_accuracy: 0.9970 - val_dense_3_accuracy: 0.9960\n",
            "Epoch 36/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0227 - dense_4_loss: 0.0142 - dense_1_loss: 0.0139 - dense_3_loss: 0.0144 - dense_4_accuracy: 0.9957 - dense_1_accuracy: 0.9960 - dense_3_accuracy: 0.9961 - val_loss: 0.0517 - val_dense_4_loss: 0.0354 - val_dense_1_loss: 0.0261 - val_dense_3_loss: 0.0280 - val_dense_4_accuracy: 0.9950 - val_dense_1_accuracy: 0.9970 - val_dense_3_accuracy: 0.9965\n",
            "Epoch 37/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0194 - dense_4_loss: 0.0121 - dense_1_loss: 0.0111 - dense_3_loss: 0.0132 - dense_4_accuracy: 0.9960 - dense_1_accuracy: 0.9964 - dense_3_accuracy: 0.9964 - val_loss: 0.0553 - val_dense_4_loss: 0.0323 - val_dense_1_loss: 0.0312 - val_dense_3_loss: 0.0455 - val_dense_4_accuracy: 0.9950 - val_dense_1_accuracy: 0.9965 - val_dense_3_accuracy: 0.9935\n",
            "Epoch 38/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0157 - dense_4_loss: 0.0093 - dense_1_loss: 0.0115 - dense_3_loss: 0.0098 - dense_4_accuracy: 0.9968 - dense_1_accuracy: 0.9966 - dense_3_accuracy: 0.9968 - val_loss: 0.0393 - val_dense_4_loss: 0.0237 - val_dense_1_loss: 0.0232 - val_dense_3_loss: 0.0287 - val_dense_4_accuracy: 0.9970 - val_dense_1_accuracy: 0.9980 - val_dense_3_accuracy: 0.9975\n",
            "Epoch 39/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0141 - dense_4_loss: 0.0081 - dense_1_loss: 0.0100 - dense_3_loss: 0.0099 - dense_4_accuracy: 0.9975 - dense_1_accuracy: 0.9969 - dense_3_accuracy: 0.9971 - val_loss: 0.0805 - val_dense_4_loss: 0.0598 - val_dense_1_loss: 0.0385 - val_dense_3_loss: 0.0304 - val_dense_4_accuracy: 0.9895 - val_dense_1_accuracy: 0.9955 - val_dense_3_accuracy: 0.9940\n",
            "Epoch 40/40\n",
            "907/907 [==============================] - 148s 163ms/step - loss: 0.0184 - dense_4_loss: 0.0113 - dense_1_loss: 0.0118 - dense_3_loss: 0.0119 - dense_4_accuracy: 0.9961 - dense_1_accuracy: 0.9961 - dense_3_accuracy: 0.9966 - val_loss: 0.0525 - val_dense_4_loss: 0.0321 - val_dense_1_loss: 0.0280 - val_dense_3_loss: 0.0402 - val_dense_4_accuracy: 0.9955 - val_dense_1_accuracy: 0.9970 - val_dense_3_accuracy: 0.9955\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 910
        },
        "id": "53UluuOE5TX_",
        "outputId": "7b339ea9-7891-4869-f5d1-d5e49ea0025a"
      },
      "source": [
        "fig, axs = plt.subplots(2, 1, figsize=(15,15))\r\n",
        "\r\n",
        "axs[0].plot(history.history['loss'])\r\n",
        "axs[0].plot(history.history['val_loss'])\r\n",
        "axs[0].title.set_text('Training Loss vs Validation Loss')\r\n",
        "axs[0].set_xlabel('Epochs')\r\n",
        "axs[0].set_ylabel('Loss')\r\n",
        "axs[0].legend(['Train','Val'])\r\n",
        "\r\n",
        "axs[1].plot(history.history['dense_4_accuracy'])\r\n",
        "axs[1].plot(history.history['val_dense_4_accuracy'])\r\n",
        "axs[1].title.set_text('Training Accuracy vs Validation Accuracy')\r\n",
        "axs[1].set_xlabel('Epochs')\r\n",
        "axs[1].set_ylabel('Accuracy')\r\n",
        "axs[1].legend(['Train', 'Val'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fd794ede278>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x1080 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iNbFiTZW5dQ7",
        "outputId": "0b733da8-860e-4640-9b55-44b93eea3b9f"
      },
      "source": [
        "model.evaluate(x_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "313/313 [==============================] - 10s 27ms/step - loss: 0.0516 - dense_4_loss: 0.0299 - dense_1_loss: 0.0335 - dense_3_loss: 0.0388 - dense_4_accuracy: 0.9929 - dense_1_accuracy: 0.9923 - dense_3_accuracy: 0.9918\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.05155260115861893,\n",
              " 0.0298592709004879,\n",
              " 0.03350009024143219,\n",
              " 0.03881095349788666,\n",
              " 0.992900013923645,\n",
              " 0.9922999739646912,\n",
              " 0.9918000102043152]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    }
  ]
 }
No results found