hw2_model4_Kfold.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "hw2_model4_Kfold.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/PriyankaGona/MachineLearning-Assignments/blob/master/hw2_model4_Kfold.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "t7njin_4BDSx",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Required headers and libraries are imported\n",
        "import keras\n",
        "from keras.datasets import cifar10\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout, Activation, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.optimizers import Adam\n",
        "from keras.losses import categorical_crossentropy\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.utils import np_utils\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.utils import shuffle"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "aNKznfiEBQe1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Required variables are declared\n",
        "batch_size = 32\n",
        "num_classes = 10\n",
        "epochs = 100\n",
        "\n",
        "validation_loss = list()\n",
        "validation_accuracy = list()\n",
        "\n",
        "#Loading and shuffling Cifar10 dataset\n",
        "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
        "\n",
        "\n",
        "\n",
        "split_index = int(0.8 * x_train.shape[0])\n",
        "\n",
        "\n",
        "y_train = keras.utils.to_categorical(y_train, num_classes)\n",
        "y_test = keras.utils.to_categorical(y_test, num_classes)\n",
        "\n",
        "x_train = x_train.astype('float32')\n",
        "x_test = x_test.astype('float32')\n",
        "\n",
        "x_train /= 255\n",
        "x_test /= 255"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "rtfiYPR-BVf6",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def getModel(input_shape):\n",
        "  \n",
        "\tmodel = Sequential()\n",
        "\n",
        "\tmodel.add(Conv2D(32, (3, 3), padding='same', input_shape=input_shape))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(Conv2D(32, (3, 3)))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "\tmodel.add(Dropout(0.25))\n",
        "\n",
        "\tmodel.add(Conv2D(64, (3, 3), padding='same'))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(Conv2D(64, (3, 3)))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "\n",
        "\tmodel.add(Dropout(0.25))\n",
        "\n",
        "\tmodel.add(Conv2D(128, (2, 2), padding='same'))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(Conv2D(128, (2, 2)))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "\n",
        "\tmodel.add(Dropout(0.25))\n",
        "\n",
        "\tmodel.add(Flatten())\n",
        "\n",
        "\tmodel.add(Dense(1024))\n",
        "\tmodel.add(Activation('relu'))\n",
        "\n",
        "\tmodel.add(Dropout(0.5))\n",
        "\n",
        "\tmodel.add(Dense(num_classes))\n",
        "\tmodel.add(Activation('softmax'))\n",
        "\t   \n",
        "\t \n",
        "\tmodel.compile(loss='categorical_crossentropy', optimizer=Adam(lr=0.0001, decay=1e-6), metrics=['accuracy'])\n",
        "\n",
        "\treturn model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "F4mD3r4nRKGK",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def train_model(model, train_x, train_y, validate_x, validate_y):\n",
        "  # Data Augmentation\n",
        "\tdatagen = ImageDataGenerator(rotation_range=0, width_shift_range=0.1,height_shift_range=0.1,fill_mode='nearest', horizontal_flip=True, vertical_flip=False, rescale=None)\n",
        "\n",
        "\tdatagen.fit(train_x)\n",
        "\n",
        "\tfor e in range(10):\n",
        "\t\tbatches = 0\n",
        "\t\tfor x_batch, y_batch in datagen.flow(train_x, train_y, batch_size=40000):\n",
        "\t\t\tmodel.fit(x_batch, y_batch)\n",
        "\t\t\tbatches += 1\n",
        "\t\t\tif batches >= 1:\n",
        "\t\t\t\t# we need to break the loop by hand because\n",
        "\t\t\t\t# the generator loops indefinitely\n",
        "\t\t\t\tbreak\n",
        "\t\t\t\t\n",
        "\t\t\t\t\n",
        "\treturn model.fit(train_x, train_y, batch_size=batch_size, epochs=epochs, verbose=1, validation_data=(validate_x, validate_y))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "BKw0KY0wIB18",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Model is tested on validation data\n",
        "def test_model(model,validate_x, validate_y):\n",
        "\t\n",
        "  score = model.evaluate(validate_x, validate_y, verbose=0)\n",
        "\t\n",
        "  print('Test loss:', score[0])\n",
        "  print('Test accuracy:', score[1])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "-WgplgzQIGZ-",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def plot_training_graph(training_history, foldNumber):\n",
        "\t\n",
        "\tplotaccuracy = plt.plot(range(1,epochs+1), training_history.history['acc'],range(1,epochs+1), training_history.history['val_acc'])\n",
        "\t\n",
        "\tplt.title(\"Fold Number \"+str(foldNumber))\n",
        "\tplt.xlabel('Number of epochs')\n",
        "\tplt.ylabel('Accuracy')\n",
        "\n",
        "\tplt.legend(('Train Accuracy','Test Accuracy'))\n",
        "\n",
        "\tplt.show(plotaccuracy)\n",
        "\n",
        "\tprint('\\n\\nvalidation loss:', training_history.history['val_loss'][-1])\n",
        "\tprint('validation accuracy:', training_history.history['val_acc'][-1])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "FqvhsWMwGAvN",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def kfold_validation(k,x_folds,y_folds):\n",
        "\n",
        "\t# Dividing data into training and validation sets\n",
        "\ttrain_x = list(x_folds)\n",
        "\tvalidate_x = train_x.pop(k-1)\n",
        "\ttrain_x = np.concatenate(train_x)\n",
        "\n",
        "\ttrain_y = list(y_folds)\n",
        "\tvalidate_y = train_y.pop(k-1)\n",
        "\ttrain_y = np.concatenate(train_y)\n",
        "\n",
        "\tprint(\"\\nnumber of samples in training set: \",len(train_x))\n",
        "\tprint(\"number of samples in validation set: \",len(validate_x))\n",
        "\n",
        "\tinput_shape = train_x.shape[1:]\n",
        "\tprint(\"Shape of images: \",input_shape)\n",
        "\n",
        "\tmodel = getModel(input_shape)\n",
        "\n",
        "\tHistory = train_model(model, train_x, train_y, validate_x, validate_y)\n",
        "\n",
        "\tvalidation_loss.append(History.history['val_loss'][-1])\n",
        "\tvalidation_accuracy.append(History.history['val_acc'][-1])\n",
        "\n",
        "\ttest_model(model,validate_x,validate_y)\n",
        "\n",
        "\tplot_training_graph(History, k)\n",
        "\n",
        "\treturn model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "NW6xrwfzv8ld",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "x_folds = np.array_split(x_train, 5)\n",
        "y_folds = np.array_split(y_train, 5)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "vpgOdmJIGGpJ",
        "colab_type": "code",
        "outputId": "08e3bf3a-667f-48c3-fb73-bb5853f09e9e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 8553
        }
      },
      "cell_type": "code",
      "source": [
        "#get the new training and validation set on fold Number 1 by calling kfold_validation\n",
        "model = kfold_validation(1,x_folds,y_folds)\n"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "40000/40000 [==============================] - 17s 433us/step - loss: 1.4833 - acc: 0.4540\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 435us/step - loss: 1.4197 - acc: 0.4836\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 437us/step - loss: 1.3659 - acc: 0.5027\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 430us/step - loss: 1.3084 - acc: 0.5271\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 432us/step - loss: 1.2714 - acc: 0.5402\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 429us/step - loss: 1.2344 - acc: 0.5570\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 441us/step - loss: 1.1947 - acc: 0.5716\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 1.0828 - acc: 0.6119 - val_loss: 0.9733 - val_acc: 0.6502\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 1.0523 - acc: 0.6251 - val_loss: 0.9699 - val_acc: 0.6505\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 1.0217 - acc: 0.6359 - val_loss: 0.9395 - val_acc: 0.6668\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 19s 485us/step - loss: 0.9935 - acc: 0.6458 - val_loss: 0.9239 - val_acc: 0.6712\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 19s 485us/step - loss: 0.9712 - acc: 0.6532 - val_loss: 0.8859 - val_acc: 0.6788\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.9511 - acc: 0.6599 - val_loss: 0.9008 - val_acc: 0.6805\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.9308 - acc: 0.6684 - val_loss: 0.8520 - val_acc: 0.6964\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.9079 - acc: 0.6786 - val_loss: 0.8457 - val_acc: 0.6977\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.8887 - acc: 0.6841 - val_loss: 0.8176 - val_acc: 0.7043\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.8746 - acc: 0.6891 - val_loss: 0.8018 - val_acc: 0.7078\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 19s 485us/step - loss: 0.8558 - acc: 0.6976 - val_loss: 0.8075 - val_acc: 0.7111\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.8383 - acc: 0.7035 - val_loss: 0.8019 - val_acc: 0.7098\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.8181 - acc: 0.7091 - val_loss: 0.7745 - val_acc: 0.7215\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.8099 - acc: 0.7127 - val_loss: 0.7575 - val_acc: 0.7295\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.7935 - acc: 0.7185 - val_loss: 0.7453 - val_acc: 0.7363\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7806 - acc: 0.7248 - val_loss: 0.7467 - val_acc: 0.7318\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7665 - acc: 0.7292 - val_loss: 0.7368 - val_acc: 0.7374\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7505 - acc: 0.7341 - val_loss: 0.7097 - val_acc: 0.7468\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7380 - acc: 0.7382 - val_loss: 0.7138 - val_acc: 0.7479\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7253 - acc: 0.7446 - val_loss: 0.7386 - val_acc: 0.7433\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.7122 - acc: 0.7476 - val_loss: 0.7359 - val_acc: 0.7456\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.7013 - acc: 0.7512 - val_loss: 0.6699 - val_acc: 0.7646\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.6891 - acc: 0.7545 - val_loss: 0.6648 - val_acc: 0.7618\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.6813 - acc: 0.7587 - val_loss: 0.6618 - val_acc: 0.7695\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6673 - acc: 0.7620 - val_loss: 0.6481 - val_acc: 0.7715\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6589 - acc: 0.7669 - val_loss: 0.6490 - val_acc: 0.7680\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.6449 - acc: 0.7735 - val_loss: 0.6443 - val_acc: 0.7741\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.6391 - acc: 0.7759 - val_loss: 0.6304 - val_acc: 0.7798\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.6304 - acc: 0.7760 - val_loss: 0.6481 - val_acc: 0.7724\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.6249 - acc: 0.7788 - val_loss: 0.6315 - val_acc: 0.7795\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.6102 - acc: 0.7837 - val_loss: 0.6225 - val_acc: 0.7844\n",
            "Epoch 32/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.6035 - acc: 0.7870 - val_loss: 0.6217 - val_acc: 0.7809\n",
            "Epoch 33/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.5978 - acc: 0.7867 - val_loss: 0.6224 - val_acc: 0.7855\n",
            "Epoch 34/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.5876 - acc: 0.7906 - val_loss: 0.6122 - val_acc: 0.7878\n",
            "Epoch 35/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.5811 - acc: 0.7944 - val_loss: 0.6166 - val_acc: 0.7862\n",
            "Epoch 36/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.5710 - acc: 0.7968 - val_loss: 0.5980 - val_acc: 0.7923\n",
            "Epoch 37/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5658 - acc: 0.7984 - val_loss: 0.5843 - val_acc: 0.7962\n",
            "Epoch 38/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5560 - acc: 0.8032 - val_loss: 0.6100 - val_acc: 0.7866\n",
            "Epoch 39/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5483 - acc: 0.8069 - val_loss: 0.5803 - val_acc: 0.7997\n",
            "Epoch 40/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5389 - acc: 0.8072 - val_loss: 0.5820 - val_acc: 0.7957\n",
            "Epoch 41/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5350 - acc: 0.8090 - val_loss: 0.5992 - val_acc: 0.7950\n",
            "Epoch 42/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5340 - acc: 0.8102 - val_loss: 0.5767 - val_acc: 0.7978\n",
            "Epoch 43/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5297 - acc: 0.8116 - val_loss: 0.5792 - val_acc: 0.8018\n",
            "Epoch 44/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5145 - acc: 0.8172 - val_loss: 0.5783 - val_acc: 0.8014\n",
            "Epoch 45/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5152 - acc: 0.8172 - val_loss: 0.5682 - val_acc: 0.8052\n",
            "Epoch 46/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.5002 - acc: 0.8232 - val_loss: 0.5653 - val_acc: 0.8064\n",
            "Epoch 47/100\n",
            "40000/40000 [==============================] - 19s 468us/step - loss: 0.5011 - acc: 0.8222 - val_loss: 0.5688 - val_acc: 0.8053\n",
            "Epoch 48/100\n",
            "40000/40000 [==============================] - 19s 468us/step - loss: 0.4944 - acc: 0.8253 - val_loss: 0.5622 - val_acc: 0.8067\n",
            "Epoch 49/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.4886 - acc: 0.8259 - val_loss: 0.5531 - val_acc: 0.8084\n",
            "Epoch 50/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.4823 - acc: 0.8280 - val_loss: 0.5585 - val_acc: 0.8073\n",
            "Epoch 51/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.4747 - acc: 0.8316 - val_loss: 0.5495 - val_acc: 0.8067\n",
            "Epoch 52/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4719 - acc: 0.8310 - val_loss: 0.5613 - val_acc: 0.8074\n",
            "Epoch 53/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4657 - acc: 0.8333 - val_loss: 0.5544 - val_acc: 0.8086\n",
            "Epoch 54/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4563 - acc: 0.8382 - val_loss: 0.5528 - val_acc: 0.8139\n",
            "Epoch 55/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.4593 - acc: 0.8400 - val_loss: 0.5437 - val_acc: 0.8127\n",
            "Epoch 56/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4536 - acc: 0.8376 - val_loss: 0.5396 - val_acc: 0.8130\n",
            "Epoch 57/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4468 - acc: 0.8417 - val_loss: 0.5444 - val_acc: 0.8099\n",
            "Epoch 58/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4440 - acc: 0.8424 - val_loss: 0.5363 - val_acc: 0.8146\n",
            "Epoch 59/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4369 - acc: 0.8445 - val_loss: 0.5399 - val_acc: 0.8106\n",
            "Epoch 60/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4279 - acc: 0.8477 - val_loss: 0.5526 - val_acc: 0.8145\n",
            "Epoch 61/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4300 - acc: 0.8474 - val_loss: 0.5466 - val_acc: 0.8141\n",
            "Epoch 62/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.4207 - acc: 0.8483 - val_loss: 0.5420 - val_acc: 0.8155\n",
            "Epoch 63/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4184 - acc: 0.8486 - val_loss: 0.5544 - val_acc: 0.8109\n",
            "Epoch 64/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4199 - acc: 0.8483 - val_loss: 0.5350 - val_acc: 0.8148\n",
            "Epoch 65/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4100 - acc: 0.8537 - val_loss: 0.5468 - val_acc: 0.8142\n",
            "Epoch 66/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4096 - acc: 0.8554 - val_loss: 0.5286 - val_acc: 0.8189\n",
            "Epoch 67/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.4038 - acc: 0.8557 - val_loss: 0.5330 - val_acc: 0.8187\n",
            "Epoch 68/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3972 - acc: 0.8595 - val_loss: 0.5397 - val_acc: 0.8148\n",
            "Epoch 69/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3930 - acc: 0.8605 - val_loss: 0.5388 - val_acc: 0.8189\n",
            "Epoch 70/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3920 - acc: 0.8608 - val_loss: 0.5266 - val_acc: 0.8213\n",
            "Epoch 71/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3833 - acc: 0.8633 - val_loss: 0.5387 - val_acc: 0.8213\n",
            "Epoch 72/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3879 - acc: 0.8610 - val_loss: 0.5391 - val_acc: 0.8142\n",
            "Epoch 73/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3843 - acc: 0.8633 - val_loss: 0.5330 - val_acc: 0.8199\n",
            "Epoch 74/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3768 - acc: 0.8652 - val_loss: 0.5212 - val_acc: 0.8257\n",
            "Epoch 75/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.3703 - acc: 0.8679 - val_loss: 0.5310 - val_acc: 0.8231\n",
            "Epoch 76/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3699 - acc: 0.8683 - val_loss: 0.5263 - val_acc: 0.8269\n",
            "Epoch 77/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3690 - acc: 0.8688 - val_loss: 0.5254 - val_acc: 0.8253\n",
            "Epoch 78/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3583 - acc: 0.8700 - val_loss: 0.5205 - val_acc: 0.8273\n",
            "Epoch 79/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.3609 - acc: 0.8707 - val_loss: 0.5246 - val_acc: 0.8255\n",
            "Epoch 80/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.3581 - acc: 0.8728 - val_loss: 0.5322 - val_acc: 0.8221\n",
            "Epoch 81/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3535 - acc: 0.8726 - val_loss: 0.5413 - val_acc: 0.8226\n",
            "Epoch 82/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3515 - acc: 0.8741 - val_loss: 0.5415 - val_acc: 0.8209\n",
            "Epoch 83/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3476 - acc: 0.8752 - val_loss: 0.5418 - val_acc: 0.8228\n",
            "Epoch 84/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.3409 - acc: 0.8766 - val_loss: 0.5510 - val_acc: 0.8218\n",
            "Epoch 85/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3471 - acc: 0.8754 - val_loss: 0.5272 - val_acc: 0.8237\n",
            "Epoch 86/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3375 - acc: 0.8798 - val_loss: 0.5355 - val_acc: 0.8263\n",
            "Epoch 87/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3330 - acc: 0.8791 - val_loss: 0.5316 - val_acc: 0.8260\n",
            "Epoch 88/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3267 - acc: 0.8822 - val_loss: 0.5316 - val_acc: 0.8244\n",
            "Epoch 89/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3350 - acc: 0.8789 - val_loss: 0.5538 - val_acc: 0.8217\n",
            "Epoch 90/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3242 - acc: 0.8829 - val_loss: 0.5372 - val_acc: 0.8245\n",
            "Epoch 91/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3247 - acc: 0.8827 - val_loss: 0.5275 - val_acc: 0.8267\n",
            "Epoch 92/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3242 - acc: 0.8831 - val_loss: 0.5414 - val_acc: 0.8239\n",
            "Epoch 93/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3205 - acc: 0.8843 - val_loss: 0.5252 - val_acc: 0.8281\n",
            "Epoch 94/100\n",
            "40000/40000 [==============================] - 19s 468us/step - loss: 0.3191 - acc: 0.8853 - val_loss: 0.5356 - val_acc: 0.8240\n",
            "Epoch 95/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3161 - acc: 0.8861 - val_loss: 0.5214 - val_acc: 0.8285\n",
            "Epoch 96/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3065 - acc: 0.8908 - val_loss: 0.5381 - val_acc: 0.8276\n",
            "Epoch 97/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3062 - acc: 0.8895 - val_loss: 0.5336 - val_acc: 0.8289\n",
            "Epoch 98/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3058 - acc: 0.8903 - val_loss: 0.5293 - val_acc: 0.8317\n",
            "Epoch 99/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.3065 - acc: 0.8913 - val_loss: 0.5211 - val_acc: 0.8303\n",
            "Epoch 100/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3012 - acc: 0.8935 - val_loss: 0.5283 - val_acc: 0.8308\n",
            "Test loss: 0.5283313675880432\n",
            "Test accuracy: 0.8308\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "validation loss: 0.5283313675880432\n",
            "validation accuracy: 0.8308\n",
            "\n",
            "number of samples in training set:  40000\n",
            "number of samples in validation set:  10000\n",
            "Shape of images:  (32, 32, 3)\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 494us/step - loss: 1.9849 - acc: 0.2498\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 451us/step - loss: 1.6869 - acc: 0.3725\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 451us/step - loss: 1.5658 - acc: 0.4254\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 445us/step - loss: 1.4877 - acc: 0.4535\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 446us/step - loss: 1.4211 - acc: 0.4826\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 438us/step - loss: 1.3621 - acc: 0.5070\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 450us/step - loss: 1.3146 - acc: 0.5248\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 464us/step - loss: 1.2756 - acc: 0.5393\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 444us/step - loss: 1.2439 - acc: 0.5558\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 454us/step - loss: 1.2011 - acc: 0.5707\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 20s 498us/step - loss: 1.0939 - acc: 0.6114 - val_loss: 0.9890 - val_acc: 0.6477\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 1.0607 - acc: 0.6238 - val_loss: 0.9732 - val_acc: 0.6535\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 1.0361 - acc: 0.6328 - val_loss: 0.9474 - val_acc: 0.6624\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 1.0071 - acc: 0.6429 - val_loss: 0.9039 - val_acc: 0.6800\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.9820 - acc: 0.6517 - val_loss: 0.8949 - val_acc: 0.6812\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.9636 - acc: 0.6584 - val_loss: 0.8872 - val_acc: 0.6866\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.9397 - acc: 0.6682 - val_loss: 0.8795 - val_acc: 0.6877\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.9221 - acc: 0.6718 - val_loss: 0.8608 - val_acc: 0.6947\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.9022 - acc: 0.6777 - val_loss: 0.8319 - val_acc: 0.7053\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.8785 - acc: 0.6901 - val_loss: 0.8564 - val_acc: 0.6941\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 20s 491us/step - loss: 0.8665 - acc: 0.6918 - val_loss: 0.8110 - val_acc: 0.7129\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.8427 - acc: 0.7036 - val_loss: 0.8342 - val_acc: 0.7094\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.8284 - acc: 0.7077 - val_loss: 0.7978 - val_acc: 0.7162\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.8201 - acc: 0.7137 - val_loss: 0.7731 - val_acc: 0.7240\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.8048 - acc: 0.7162 - val_loss: 0.7677 - val_acc: 0.7316\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7897 - acc: 0.7211 - val_loss: 0.7882 - val_acc: 0.7204\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7807 - acc: 0.7233 - val_loss: 0.7493 - val_acc: 0.7338\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7705 - acc: 0.7251 - val_loss: 0.7318 - val_acc: 0.7429\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.7515 - acc: 0.7330 - val_loss: 0.7583 - val_acc: 0.7314\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.7423 - acc: 0.7358 - val_loss: 0.7157 - val_acc: 0.7457\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7295 - acc: 0.7437 - val_loss: 0.7240 - val_acc: 0.7474\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.7167 - acc: 0.7449 - val_loss: 0.7068 - val_acc: 0.7482\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.7100 - acc: 0.7510 - val_loss: 0.6985 - val_acc: 0.7502\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.6911 - acc: 0.7560 - val_loss: 0.6870 - val_acc: 0.7571\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.6871 - acc: 0.7570 - val_loss: 0.7302 - val_acc: 0.7415\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.6759 - acc: 0.7617 - val_loss: 0.7004 - val_acc: 0.7568\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.6687 - acc: 0.7613 - val_loss: 0.6903 - val_acc: 0.7558\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.6572 - acc: 0.7685 - val_loss: 0.6736 - val_acc: 0.7639\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.6540 - acc: 0.7678 - val_loss: 0.6660 - val_acc: 0.7645\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.6413 - acc: 0.7714 - val_loss: 0.6668 - val_acc: 0.7633\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.6375 - acc: 0.7744 - val_loss: 0.6767 - val_acc: 0.7613\n",
            "Epoch 32/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.6259 - acc: 0.7784 - val_loss: 0.6565 - val_acc: 0.7698\n",
            "Epoch 33/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.6176 - acc: 0.7829 - val_loss: 0.6535 - val_acc: 0.7717\n",
            "Epoch 34/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.6076 - acc: 0.7862 - val_loss: 0.6335 - val_acc: 0.7757\n",
            "Epoch 35/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.5962 - acc: 0.7886 - val_loss: 0.6322 - val_acc: 0.7759\n",
            "Epoch 36/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.5898 - acc: 0.7898 - val_loss: 0.6355 - val_acc: 0.7767\n",
            "Epoch 37/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.5845 - acc: 0.7922 - val_loss: 0.6165 - val_acc: 0.7833\n",
            "Epoch 38/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.5776 - acc: 0.7962 - val_loss: 0.6148 - val_acc: 0.7863\n",
            "Epoch 39/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.5698 - acc: 0.7971 - val_loss: 0.6180 - val_acc: 0.7879\n",
            "Epoch 40/100\n",
            "40000/40000 [==============================] - 20s 491us/step - loss: 0.5664 - acc: 0.7994 - val_loss: 0.6249 - val_acc: 0.7801\n",
            "Epoch 41/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.5562 - acc: 0.8030 - val_loss: 0.6068 - val_acc: 0.7901\n",
            "Epoch 42/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.5493 - acc: 0.8044 - val_loss: 0.6051 - val_acc: 0.7876\n",
            "Epoch 43/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.5432 - acc: 0.8090 - val_loss: 0.5949 - val_acc: 0.7925\n",
            "Epoch 44/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5385 - acc: 0.8080 - val_loss: 0.5973 - val_acc: 0.7907\n",
            "Epoch 45/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.5295 - acc: 0.8117 - val_loss: 0.5957 - val_acc: 0.7915\n",
            "Epoch 46/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.5253 - acc: 0.8140 - val_loss: 0.5861 - val_acc: 0.7946\n",
            "Epoch 47/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.5167 - acc: 0.8177 - val_loss: 0.5858 - val_acc: 0.7954\n",
            "Epoch 48/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5142 - acc: 0.8189 - val_loss: 0.6020 - val_acc: 0.7918\n",
            "Epoch 49/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.5068 - acc: 0.8181 - val_loss: 0.5891 - val_acc: 0.7962\n",
            "Epoch 50/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.5002 - acc: 0.8223 - val_loss: 0.5925 - val_acc: 0.7941\n",
            "Epoch 51/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4956 - acc: 0.8242 - val_loss: 0.5864 - val_acc: 0.7965\n",
            "Epoch 52/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4897 - acc: 0.8265 - val_loss: 0.5894 - val_acc: 0.7958\n",
            "Epoch 53/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4814 - acc: 0.8295 - val_loss: 0.5683 - val_acc: 0.8009\n",
            "Epoch 54/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4763 - acc: 0.8307 - val_loss: 0.5881 - val_acc: 0.7968\n",
            "Epoch 55/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.4761 - acc: 0.8300 - val_loss: 0.5674 - val_acc: 0.8007\n",
            "Epoch 56/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.4735 - acc: 0.8319 - val_loss: 0.5680 - val_acc: 0.8055\n",
            "Epoch 57/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4672 - acc: 0.8350 - val_loss: 0.5773 - val_acc: 0.8018\n",
            "Epoch 58/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.4613 - acc: 0.8371 - val_loss: 0.5666 - val_acc: 0.8040\n",
            "Epoch 59/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4553 - acc: 0.8387 - val_loss: 0.5642 - val_acc: 0.8053\n",
            "Epoch 60/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4504 - acc: 0.8392 - val_loss: 0.5681 - val_acc: 0.8022\n",
            "Epoch 61/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.4480 - acc: 0.8405 - val_loss: 0.5549 - val_acc: 0.8051\n",
            "Epoch 62/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4420 - acc: 0.8416 - val_loss: 0.5553 - val_acc: 0.8055\n",
            "Epoch 63/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.4407 - acc: 0.8435 - val_loss: 0.5613 - val_acc: 0.8054\n",
            "Epoch 64/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4400 - acc: 0.8441 - val_loss: 0.5659 - val_acc: 0.8032\n",
            "Epoch 65/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.4246 - acc: 0.8473 - val_loss: 0.5643 - val_acc: 0.8060\n",
            "Epoch 66/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.4299 - acc: 0.8477 - val_loss: 0.5559 - val_acc: 0.8113\n",
            "Epoch 67/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.4205 - acc: 0.8482 - val_loss: 0.5613 - val_acc: 0.8103\n",
            "Epoch 68/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.4211 - acc: 0.8503 - val_loss: 0.5541 - val_acc: 0.8117\n",
            "Epoch 69/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4157 - acc: 0.8512 - val_loss: 0.5528 - val_acc: 0.8098\n",
            "Epoch 70/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.4085 - acc: 0.8541 - val_loss: 0.5701 - val_acc: 0.8060\n",
            "Epoch 71/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4074 - acc: 0.8528 - val_loss: 0.5581 - val_acc: 0.8095\n",
            "Epoch 72/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4051 - acc: 0.8558 - val_loss: 0.5557 - val_acc: 0.8116\n",
            "Epoch 73/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.3957 - acc: 0.8586 - val_loss: 0.5593 - val_acc: 0.8120\n",
            "Epoch 74/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3965 - acc: 0.8582 - val_loss: 0.5417 - val_acc: 0.8168\n",
            "Epoch 75/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3920 - acc: 0.8580 - val_loss: 0.5647 - val_acc: 0.8091\n",
            "Epoch 76/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3834 - acc: 0.8626 - val_loss: 0.5709 - val_acc: 0.8093\n",
            "Epoch 77/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3838 - acc: 0.8630 - val_loss: 0.5579 - val_acc: 0.8109\n",
            "Epoch 78/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3858 - acc: 0.8622 - val_loss: 0.5593 - val_acc: 0.8152\n",
            "Epoch 79/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3729 - acc: 0.8647 - val_loss: 0.5593 - val_acc: 0.8132\n",
            "Epoch 80/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3733 - acc: 0.8653 - val_loss: 0.5431 - val_acc: 0.8196\n",
            "Epoch 81/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3762 - acc: 0.8651 - val_loss: 0.5449 - val_acc: 0.8129\n",
            "Epoch 82/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3630 - acc: 0.8694 - val_loss: 0.5505 - val_acc: 0.8151\n",
            "Epoch 83/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3685 - acc: 0.8671 - val_loss: 0.5701 - val_acc: 0.8128\n",
            "Epoch 84/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3598 - acc: 0.8707 - val_loss: 0.5640 - val_acc: 0.8145\n",
            "Epoch 85/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3625 - acc: 0.8705 - val_loss: 0.5589 - val_acc: 0.8157\n",
            "Epoch 86/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3552 - acc: 0.8730 - val_loss: 0.5528 - val_acc: 0.8159\n",
            "Epoch 87/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3535 - acc: 0.8713 - val_loss: 0.5645 - val_acc: 0.8141\n",
            "Epoch 88/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3516 - acc: 0.8740 - val_loss: 0.5488 - val_acc: 0.8167\n",
            "Epoch 89/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3507 - acc: 0.8742 - val_loss: 0.5343 - val_acc: 0.8205\n",
            "Epoch 90/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3468 - acc: 0.8742 - val_loss: 0.5380 - val_acc: 0.8219\n",
            "Epoch 91/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3401 - acc: 0.8785 - val_loss: 0.5395 - val_acc: 0.8207\n",
            "Epoch 92/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.3339 - acc: 0.8819 - val_loss: 0.5465 - val_acc: 0.8229\n",
            "Epoch 93/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3325 - acc: 0.8790 - val_loss: 0.5511 - val_acc: 0.8181\n",
            "Epoch 94/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3320 - acc: 0.8800 - val_loss: 0.5456 - val_acc: 0.8196\n",
            "Epoch 95/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3279 - acc: 0.8821 - val_loss: 0.5450 - val_acc: 0.8212\n",
            "Epoch 96/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3267 - acc: 0.8826 - val_loss: 0.5371 - val_acc: 0.8216\n",
            "Epoch 97/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3278 - acc: 0.8816 - val_loss: 0.5426 - val_acc: 0.8252\n",
            "Epoch 98/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3287 - acc: 0.8813 - val_loss: 0.5576 - val_acc: 0.8182\n",
            "Epoch 99/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3201 - acc: 0.8852 - val_loss: 0.5489 - val_acc: 0.8206\n",
            "Epoch 100/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3244 - acc: 0.8851 - val_loss: 0.5465 - val_acc: 0.8225\n",
            "Test loss: 0.5464786044120788\n",
            "Test accuracy: 0.8225\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "validation loss: 0.5464786044120788\n",
            "validation accuracy: 0.8225\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "l0AR9hweINN0",
        "colab_type": "code",
        "outputId": "69b5b029-ef51-42c7-988a-e1ecc15c8c5a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1560
        }
      },
      "cell_type": "code",
      "source": [
        "#get the new training and validation set on fold Number 2 by calling kfold_validation\n",
        "model = kfold_validation(2,x_folds,y_folds)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "number of samples in training set:  40000\n",
            "number of samples in validation set:  10000\n",
            "Shape of images:  (32, 32, 3)\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 493us/step - loss: 1.9762 - acc: 0.2545\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 455us/step - loss: 1.6805 - acc: 0.3756\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 448us/step - loss: 1.5583 - acc: 0.4234\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 452us/step - loss: 1.4778 - acc: 0.4579\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 453us/step - loss: 1.4098 - acc: 0.4846\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 453us/step - loss: 1.3514 - acc: 0.5108\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 454us/step - loss: 1.3037 - acc: 0.5286\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 454us/step - loss: 1.2571 - acc: 0.5471\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 447us/step - loss: 1.2202 - acc: 0.5630\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 451us/step - loss: 1.1880 - acc: 0.5761\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 20s 498us/step - loss: 1.0770 - acc: 0.6148 - val_loss: 0.9776 - val_acc: 0.6512\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 20s 492us/step - loss: 1.0362 - acc: 0.6309 - val_loss: 0.9726 - val_acc: 0.6533\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 20s 491us/step - loss: 1.0066 - acc: 0.6420 - val_loss: 0.9269 - val_acc: 0.6694\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.9799 - acc: 0.6520 - val_loss: 0.8955 - val_acc: 0.6747\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.9532 - acc: 0.6625 - val_loss: 0.8881 - val_acc: 0.6786\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.9256 - acc: 0.6707 - val_loss: 0.8622 - val_acc: 0.6923\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.9055 - acc: 0.6792 - val_loss: 0.8579 - val_acc: 0.6974\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.8844 - acc: 0.6866 - val_loss: 0.8111 - val_acc: 0.7125\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.8658 - acc: 0.6952 - val_loss: 0.8090 - val_acc: 0.7111\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.8469 - acc: 0.7014 - val_loss: 0.8301 - val_acc: 0.7026\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.8287 - acc: 0.7094 - val_loss: 0.7776 - val_acc: 0.7243\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.8080 - acc: 0.7142 - val_loss: 0.7529 - val_acc: 0.7307\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.7912 - acc: 0.7214 - val_loss: 0.7449 - val_acc: 0.7339\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7770 - acc: 0.7246 - val_loss: 0.7525 - val_acc: 0.7340\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7593 - acc: 0.7324 - val_loss: 0.7320 - val_acc: 0.7406\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.7463 - acc: 0.7363 - val_loss: 0.7065 - val_acc: 0.7479\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.7323 - acc: 0.7433 - val_loss: 0.7324 - val_acc: 0.7399\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7205 - acc: 0.7452 - val_loss: 0.6997 - val_acc: 0.7485\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.7037 - acc: 0.7518 - val_loss: 0.7254 - val_acc: 0.7413\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.6957 - acc: 0.7555 - val_loss: 0.6829 - val_acc: 0.7571\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.6807 - acc: 0.7606 - val_loss: 0.6813 - val_acc: 0.7556\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 20s 491us/step - loss: 0.6722 - acc: 0.7641 - val_loss: 0.6774 - val_acc: 0.7602\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 23s 584us/step - loss: 0.6623 - acc: 0.7658 - val_loss: 0.6671 - val_acc: 0.7624\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 20s 510us/step - loss: 0.6534 - acc: 0.7707 - val_loss: 0.6552 - val_acc: 0.7695\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 20s 506us/step - loss: 0.6435 - acc: 0.7753 - val_loss: 0.6432 - val_acc: 0.7706\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 20s 505us/step - loss: 0.6307 - acc: 0.7781 - val_loss: 0.6446 - val_acc: 0.7753\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 20s 505us/step - loss: 0.6256 - acc: 0.7767 - val_loss: 0.6274 - val_acc: 0.7782\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 20s 502us/step - loss: 0.6176 - acc: 0.7813 - val_loss: 0.6260 - val_acc: 0.7816\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 20s 505us/step - loss: 0.6073 - acc: 0.7877 - val_loss: 0.6289 - val_acc: 0.7820\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 20s 498us/step - loss: 0.5995 - acc: 0.7903 - val_loss: 0.6232 - val_acc: 0.7817\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 20s 496us/step - loss: 0.5981 - acc: 0.7896 - val_loss: 0.6179 - val_acc: 0.7799\n",
            "Epoch 32/100\n",
            "30240/40000 [=====================>........] - ETA: 4s - loss: 0.5749 - acc: 0.7981"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "foOilOx-VHSL",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 4381
        },
        "outputId": "afef80cf-5ba6-4702-87a7-74682e5e1153"
      },
      "cell_type": "code",
      "source": [
        "#get the new training and validation set on fold Number 3 by calling kfold_validation\n",
        "model = kfold_validation(3,x_folds,y_folds)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "number of samples in training set:  40000\n",
            "number of samples in validation set:  10000\n",
            "Shape of images:  (32, 32, 3)\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 466us/step - loss: 1.9682 - acc: 0.2600\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 440us/step - loss: 1.6752 - acc: 0.3740\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 435us/step - loss: 1.5620 - acc: 0.4243\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 440us/step - loss: 1.4739 - acc: 0.4597\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 1.4125 - acc: 0.4853\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 443us/step - loss: 1.3618 - acc: 0.5046\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 440us/step - loss: 1.3121 - acc: 0.5286\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 445us/step - loss: 1.2744 - acc: 0.5419\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 435us/step - loss: 1.2398 - acc: 0.5553\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 502us/step - loss: 1.2116 - acc: 0.5672\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 20s 496us/step - loss: 1.0960 - acc: 0.6092 - val_loss: 0.9820 - val_acc: 0.6574\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 1.0568 - acc: 0.6230 - val_loss: 1.0015 - val_acc: 0.6490\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 1.0309 - acc: 0.6357 - val_loss: 0.9720 - val_acc: 0.6555\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.9952 - acc: 0.6475 - val_loss: 0.9413 - val_acc: 0.6677\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 19s 485us/step - loss: 0.9749 - acc: 0.6535 - val_loss: 0.8671 - val_acc: 0.6954\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 19s 485us/step - loss: 0.9512 - acc: 0.6635 - val_loss: 0.8508 - val_acc: 0.7031\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.9316 - acc: 0.6695 - val_loss: 0.8888 - val_acc: 0.6824\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.9052 - acc: 0.6813 - val_loss: 0.8299 - val_acc: 0.7093\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.8869 - acc: 0.6877 - val_loss: 0.8257 - val_acc: 0.7100\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.8714 - acc: 0.6944 - val_loss: 0.7879 - val_acc: 0.7250\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.8503 - acc: 0.7003 - val_loss: 0.7649 - val_acc: 0.7349\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.8329 - acc: 0.7059 - val_loss: 0.7540 - val_acc: 0.7376\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.8153 - acc: 0.7119 - val_loss: 0.7301 - val_acc: 0.7472\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.7988 - acc: 0.7159 - val_loss: 0.7275 - val_acc: 0.7455\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.7856 - acc: 0.7249 - val_loss: 0.7065 - val_acc: 0.7552\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.7671 - acc: 0.7279 - val_loss: 0.6979 - val_acc: 0.7573\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.7546 - acc: 0.7332 - val_loss: 0.7000 - val_acc: 0.7525\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.7436 - acc: 0.7373 - val_loss: 0.6797 - val_acc: 0.7656\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.7291 - acc: 0.7420 - val_loss: 0.6770 - val_acc: 0.7650\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.7155 - acc: 0.7477 - val_loss: 0.6941 - val_acc: 0.7553\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.7019 - acc: 0.7536 - val_loss: 0.6540 - val_acc: 0.7695\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.6962 - acc: 0.7545 - val_loss: 0.6468 - val_acc: 0.7749\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.6796 - acc: 0.7595 - val_loss: 0.6702 - val_acc: 0.7673\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.6673 - acc: 0.7650 - val_loss: 0.6366 - val_acc: 0.7810\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.6586 - acc: 0.7668 - val_loss: 0.6541 - val_acc: 0.7722\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.6487 - acc: 0.7710 - val_loss: 0.6341 - val_acc: 0.7773\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.6425 - acc: 0.7733 - val_loss: 0.6309 - val_acc: 0.7813\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.6321 - acc: 0.7770 - val_loss: 0.6300 - val_acc: 0.7802\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.6178 - acc: 0.7804 - val_loss: 0.6289 - val_acc: 0.7859\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.6124 - acc: 0.7831 - val_loss: 0.6021 - val_acc: 0.7916\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.6031 - acc: 0.7862 - val_loss: 0.5969 - val_acc: 0.7963\n",
            "Epoch 32/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.5972 - acc: 0.7902 - val_loss: 0.5970 - val_acc: 0.7933\n",
            "Epoch 33/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.5940 - acc: 0.7880 - val_loss: 0.5878 - val_acc: 0.7978\n",
            "Epoch 34/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.5798 - acc: 0.7964 - val_loss: 0.6069 - val_acc: 0.7892\n",
            "Epoch 35/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.5697 - acc: 0.7991 - val_loss: 0.5842 - val_acc: 0.7985\n",
            "Epoch 36/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.5612 - acc: 0.8044 - val_loss: 0.5728 - val_acc: 0.8033\n",
            "Epoch 37/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.5589 - acc: 0.8014 - val_loss: 0.5988 - val_acc: 0.7944\n",
            "Epoch 38/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.5550 - acc: 0.8037 - val_loss: 0.5781 - val_acc: 0.8003\n",
            "Epoch 39/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.5470 - acc: 0.8054 - val_loss: 0.5678 - val_acc: 0.8060\n",
            "Epoch 40/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5403 - acc: 0.8091 - val_loss: 0.5743 - val_acc: 0.8034\n",
            "Epoch 41/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.5274 - acc: 0.8130 - val_loss: 0.5717 - val_acc: 0.8030\n",
            "Epoch 42/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5248 - acc: 0.8143 - val_loss: 0.5617 - val_acc: 0.8080\n",
            "Epoch 43/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5151 - acc: 0.8176 - val_loss: 0.5615 - val_acc: 0.8042\n",
            "Epoch 44/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.5143 - acc: 0.8168 - val_loss: 0.5664 - val_acc: 0.8070\n",
            "Epoch 45/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5116 - acc: 0.8201 - val_loss: 0.5516 - val_acc: 0.8130\n",
            "Epoch 46/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.5064 - acc: 0.8210 - val_loss: 0.5795 - val_acc: 0.8024\n",
            "Epoch 47/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4958 - acc: 0.8242 - val_loss: 0.5534 - val_acc: 0.8101\n",
            "Epoch 48/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4925 - acc: 0.8253 - val_loss: 0.5496 - val_acc: 0.8163\n",
            "Epoch 49/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.4846 - acc: 0.8294 - val_loss: 0.5446 - val_acc: 0.8154\n",
            "Epoch 50/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4807 - acc: 0.8289 - val_loss: 0.5398 - val_acc: 0.8140\n",
            "Epoch 51/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.4784 - acc: 0.8301 - val_loss: 0.5421 - val_acc: 0.8111\n",
            "Epoch 52/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4654 - acc: 0.8358 - val_loss: 0.5472 - val_acc: 0.8090\n",
            "Epoch 53/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.4656 - acc: 0.8348 - val_loss: 0.5399 - val_acc: 0.8142\n",
            "Epoch 54/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4642 - acc: 0.8364 - val_loss: 0.5536 - val_acc: 0.8101\n",
            "Epoch 55/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4560 - acc: 0.8371 - val_loss: 0.5541 - val_acc: 0.8109\n",
            "Epoch 56/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4480 - acc: 0.8412 - val_loss: 0.5314 - val_acc: 0.8126\n",
            "Epoch 57/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.4478 - acc: 0.8408 - val_loss: 0.5294 - val_acc: 0.8220\n",
            "Epoch 58/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.4371 - acc: 0.8433 - val_loss: 0.5387 - val_acc: 0.8186\n",
            "Epoch 59/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4342 - acc: 0.8460 - val_loss: 0.5346 - val_acc: 0.8157\n",
            "Epoch 60/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.4359 - acc: 0.8432 - val_loss: 0.5345 - val_acc: 0.8162\n",
            "Epoch 61/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.4229 - acc: 0.8495 - val_loss: 0.5337 - val_acc: 0.8163\n",
            "Epoch 62/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4232 - acc: 0.8499 - val_loss: 0.5303 - val_acc: 0.8196\n",
            "Epoch 63/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4157 - acc: 0.8544 - val_loss: 0.5223 - val_acc: 0.8219\n",
            "Epoch 64/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.4135 - acc: 0.8535 - val_loss: 0.5319 - val_acc: 0.8207\n",
            "Epoch 65/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4089 - acc: 0.8541 - val_loss: 0.5276 - val_acc: 0.8197\n",
            "Epoch 66/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4052 - acc: 0.8558 - val_loss: 0.5243 - val_acc: 0.8217\n",
            "Epoch 67/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4014 - acc: 0.8590 - val_loss: 0.5218 - val_acc: 0.8259\n",
            "Epoch 68/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3963 - acc: 0.8573 - val_loss: 0.5152 - val_acc: 0.8241\n",
            "Epoch 69/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3912 - acc: 0.8593 - val_loss: 0.5206 - val_acc: 0.8270\n",
            "Epoch 70/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3905 - acc: 0.8602 - val_loss: 0.5252 - val_acc: 0.8219\n",
            "Epoch 71/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3858 - acc: 0.8606 - val_loss: 0.5333 - val_acc: 0.8196\n",
            "Epoch 72/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3857 - acc: 0.8613 - val_loss: 0.5185 - val_acc: 0.8244\n",
            "Epoch 73/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3809 - acc: 0.8638 - val_loss: 0.5262 - val_acc: 0.8230\n",
            "Epoch 74/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3762 - acc: 0.8665 - val_loss: 0.5256 - val_acc: 0.8214\n",
            "Epoch 75/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3736 - acc: 0.8666 - val_loss: 0.5271 - val_acc: 0.8219\n",
            "Epoch 76/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3679 - acc: 0.8682 - val_loss: 0.5099 - val_acc: 0.8288\n",
            "Epoch 77/100\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 0.3637 - acc: 0.8708 - val_loss: 0.5428 - val_acc: 0.8232\n",
            "Epoch 78/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3649 - acc: 0.8702 - val_loss: 0.5133 - val_acc: 0.8286\n",
            "Epoch 79/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3599 - acc: 0.8712 - val_loss: 0.5241 - val_acc: 0.8265\n",
            "Epoch 80/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3516 - acc: 0.8744 - val_loss: 0.5322 - val_acc: 0.8231\n",
            "Epoch 81/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3524 - acc: 0.8740 - val_loss: 0.5195 - val_acc: 0.8263\n",
            "Epoch 82/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3486 - acc: 0.8751 - val_loss: 0.5126 - val_acc: 0.8294\n",
            "Epoch 83/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3493 - acc: 0.8754 - val_loss: 0.5151 - val_acc: 0.8286\n",
            "Epoch 84/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3455 - acc: 0.8769 - val_loss: 0.5170 - val_acc: 0.8295\n",
            "Epoch 85/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3441 - acc: 0.8744 - val_loss: 0.5215 - val_acc: 0.8251\n",
            "Epoch 86/100\n",
            "40000/40000 [==============================] - 19s 470us/step - loss: 0.3410 - acc: 0.8773 - val_loss: 0.5165 - val_acc: 0.8272\n",
            "Epoch 87/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3442 - acc: 0.8764 - val_loss: 0.5124 - val_acc: 0.8277\n",
            "Epoch 88/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3365 - acc: 0.8807 - val_loss: 0.5277 - val_acc: 0.8254\n",
            "Epoch 89/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3323 - acc: 0.8821 - val_loss: 0.5150 - val_acc: 0.8286\n",
            "Epoch 90/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3266 - acc: 0.8829 - val_loss: 0.5112 - val_acc: 0.8305\n",
            "Epoch 91/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3253 - acc: 0.8813 - val_loss: 0.5125 - val_acc: 0.8301\n",
            "Epoch 92/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3179 - acc: 0.8864 - val_loss: 0.5136 - val_acc: 0.8264\n",
            "Epoch 93/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3204 - acc: 0.8852 - val_loss: 0.5120 - val_acc: 0.8313\n",
            "Epoch 94/100\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 0.3175 - acc: 0.8869 - val_loss: 0.5242 - val_acc: 0.8266\n",
            "Epoch 95/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.3193 - acc: 0.8851 - val_loss: 0.5137 - val_acc: 0.8321\n",
            "Epoch 96/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3091 - acc: 0.8901 - val_loss: 0.5206 - val_acc: 0.8288\n",
            "Epoch 97/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.3109 - acc: 0.8884 - val_loss: 0.5153 - val_acc: 0.8301\n",
            "Epoch 98/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3068 - acc: 0.8886 - val_loss: 0.5209 - val_acc: 0.8265\n",
            "Epoch 99/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.3052 - acc: 0.8912 - val_loss: 0.5292 - val_acc: 0.8270\n",
            "Epoch 100/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3040 - acc: 0.8904 - val_loss: 0.5229 - val_acc: 0.8296\n",
            "Test loss: 0.5229443846702576\n",
            "Test accuracy: 0.8296\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBEOdiIiol2CoExER9RIMdSIiol7i/wNaIqY14vk23AAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "validation loss: 0.5229443846702576\n",
            "validation accuracy: 0.8296\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "y7DomGmHVJI-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 4381
        },
        "outputId": "01cec9ba-e2cd-4f64-94d2-536002f54283"
      },
      "cell_type": "code",
      "source": [
        "#get the new training and validation set on fold Number 4 by calling kfold_validation\n",
        "model = kfold_validation(4,x_folds,y_folds)"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "number of samples in training set:  40000\n",
            "number of samples in validation set:  10000\n",
            "Shape of images:  (32, 32, 3)\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 465us/step - loss: 1.9324 - acc: 0.2646\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 449us/step - loss: 1.6650 - acc: 0.3789\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 442us/step - loss: 1.5569 - acc: 0.4218\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 445us/step - loss: 1.4828 - acc: 0.4511\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 1.4192 - acc: 0.4810\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 437us/step - loss: 1.3659 - acc: 0.5051\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 439us/step - loss: 1.3158 - acc: 0.5230\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 443us/step - loss: 1.2677 - acc: 0.5438\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 433us/step - loss: 1.2350 - acc: 0.5546\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 500us/step - loss: 1.2047 - acc: 0.5687\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 23s 580us/step - loss: 1.0874 - acc: 0.6094 - val_loss: 1.0186 - val_acc: 0.6377\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 20s 507us/step - loss: 1.0513 - acc: 0.6235 - val_loss: 0.9922 - val_acc: 0.6497\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 20s 500us/step - loss: 1.0256 - acc: 0.6347 - val_loss: 0.9878 - val_acc: 0.6532\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.9922 - acc: 0.6461 - val_loss: 0.9216 - val_acc: 0.6726\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.9642 - acc: 0.6559 - val_loss: 0.9239 - val_acc: 0.6681\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.9355 - acc: 0.6678 - val_loss: 0.8899 - val_acc: 0.6833\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.9145 - acc: 0.6742 - val_loss: 0.8903 - val_acc: 0.6875\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.8962 - acc: 0.6822 - val_loss: 0.8559 - val_acc: 0.6951\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.8701 - acc: 0.6937 - val_loss: 0.8212 - val_acc: 0.7172\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.8500 - acc: 0.6999 - val_loss: 0.8362 - val_acc: 0.7045\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.8351 - acc: 0.7053 - val_loss: 0.7839 - val_acc: 0.7252\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.8150 - acc: 0.7128 - val_loss: 0.7849 - val_acc: 0.7227\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.7932 - acc: 0.7201 - val_loss: 0.7520 - val_acc: 0.7375\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.7769 - acc: 0.7258 - val_loss: 0.7435 - val_acc: 0.7359\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.7606 - acc: 0.7333 - val_loss: 0.7467 - val_acc: 0.7371\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7419 - acc: 0.7373 - val_loss: 0.7254 - val_acc: 0.7470\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7318 - acc: 0.7420 - val_loss: 0.7216 - val_acc: 0.7445\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7212 - acc: 0.7451 - val_loss: 0.6996 - val_acc: 0.7551\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.7063 - acc: 0.7507 - val_loss: 0.6877 - val_acc: 0.7566\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6923 - acc: 0.7567 - val_loss: 0.6935 - val_acc: 0.7545\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6777 - acc: 0.7634 - val_loss: 0.6814 - val_acc: 0.7616\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.6690 - acc: 0.7653 - val_loss: 0.6750 - val_acc: 0.7663\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6620 - acc: 0.7661 - val_loss: 0.6736 - val_acc: 0.7635\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6474 - acc: 0.7718 - val_loss: 0.6570 - val_acc: 0.7698\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.6420 - acc: 0.7738 - val_loss: 0.6580 - val_acc: 0.7703\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 19s 485us/step - loss: 0.6308 - acc: 0.7782 - val_loss: 0.6659 - val_acc: 0.7654\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.6239 - acc: 0.7799 - val_loss: 0.6381 - val_acc: 0.7788\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.6122 - acc: 0.7858 - val_loss: 0.6499 - val_acc: 0.7740\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.6046 - acc: 0.7865 - val_loss: 0.6381 - val_acc: 0.7768\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.5978 - acc: 0.7884 - val_loss: 0.6269 - val_acc: 0.7782\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5854 - acc: 0.7959 - val_loss: 0.6449 - val_acc: 0.7782\n",
            "Epoch 32/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5776 - acc: 0.7965 - val_loss: 0.6245 - val_acc: 0.7815\n",
            "Epoch 33/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.5724 - acc: 0.7977 - val_loss: 0.6397 - val_acc: 0.7782\n",
            "Epoch 34/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5620 - acc: 0.8002 - val_loss: 0.6209 - val_acc: 0.7850\n",
            "Epoch 35/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5614 - acc: 0.8022 - val_loss: 0.6125 - val_acc: 0.7893\n",
            "Epoch 36/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.5487 - acc: 0.8068 - val_loss: 0.6001 - val_acc: 0.7935\n",
            "Epoch 37/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.5485 - acc: 0.8082 - val_loss: 0.5957 - val_acc: 0.7929\n",
            "Epoch 38/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5343 - acc: 0.8114 - val_loss: 0.5904 - val_acc: 0.7952\n",
            "Epoch 39/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5328 - acc: 0.8127 - val_loss: 0.6018 - val_acc: 0.7915\n",
            "Epoch 40/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.5271 - acc: 0.8139 - val_loss: 0.6003 - val_acc: 0.7895\n",
            "Epoch 41/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.5188 - acc: 0.8175 - val_loss: 0.5964 - val_acc: 0.7928\n",
            "Epoch 42/100\n",
            "40000/40000 [==============================] - 19s 483us/step - loss: 0.5095 - acc: 0.8201 - val_loss: 0.5931 - val_acc: 0.7959\n",
            "Epoch 43/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.5093 - acc: 0.8192 - val_loss: 0.5809 - val_acc: 0.8001\n",
            "Epoch 44/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4960 - acc: 0.8237 - val_loss: 0.5782 - val_acc: 0.8037\n",
            "Epoch 45/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.4976 - acc: 0.8242 - val_loss: 0.5810 - val_acc: 0.8016\n",
            "Epoch 46/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4917 - acc: 0.8261 - val_loss: 0.5717 - val_acc: 0.8002\n",
            "Epoch 47/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.4833 - acc: 0.8282 - val_loss: 0.5798 - val_acc: 0.7997\n",
            "Epoch 48/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.4783 - acc: 0.8314 - val_loss: 0.5792 - val_acc: 0.8008\n",
            "Epoch 49/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.4739 - acc: 0.8310 - val_loss: 0.5647 - val_acc: 0.8058\n",
            "Epoch 50/100\n",
            "40000/40000 [==============================] - 19s 484us/step - loss: 0.4683 - acc: 0.8318 - val_loss: 0.5853 - val_acc: 0.7994\n",
            "Epoch 51/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.4586 - acc: 0.8370 - val_loss: 0.5667 - val_acc: 0.8072\n",
            "Epoch 52/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.4534 - acc: 0.8389 - val_loss: 0.5855 - val_acc: 0.8026\n",
            "Epoch 53/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.4509 - acc: 0.8392 - val_loss: 0.5870 - val_acc: 0.8018\n",
            "Epoch 54/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.4467 - acc: 0.8430 - val_loss: 0.5738 - val_acc: 0.8081\n",
            "Epoch 55/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.4423 - acc: 0.8431 - val_loss: 0.5661 - val_acc: 0.8085\n",
            "Epoch 56/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.4382 - acc: 0.8437 - val_loss: 0.5622 - val_acc: 0.8055\n",
            "Epoch 57/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.4342 - acc: 0.8455 - val_loss: 0.5490 - val_acc: 0.8148\n",
            "Epoch 58/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4311 - acc: 0.8452 - val_loss: 0.5576 - val_acc: 0.8154\n",
            "Epoch 59/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.4255 - acc: 0.8454 - val_loss: 0.5600 - val_acc: 0.8112\n",
            "Epoch 60/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4202 - acc: 0.8510 - val_loss: 0.5529 - val_acc: 0.8177\n",
            "Epoch 61/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4171 - acc: 0.8520 - val_loss: 0.5686 - val_acc: 0.8134\n",
            "Epoch 62/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4111 - acc: 0.8549 - val_loss: 0.5620 - val_acc: 0.8130\n",
            "Epoch 63/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4111 - acc: 0.8541 - val_loss: 0.5508 - val_acc: 0.8166\n",
            "Epoch 64/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4056 - acc: 0.8560 - val_loss: 0.5720 - val_acc: 0.8088\n",
            "Epoch 65/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3978 - acc: 0.8588 - val_loss: 0.5492 - val_acc: 0.8201\n",
            "Epoch 66/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3946 - acc: 0.8568 - val_loss: 0.5562 - val_acc: 0.8163\n",
            "Epoch 67/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3877 - acc: 0.8595 - val_loss: 0.5442 - val_acc: 0.8169\n",
            "Epoch 68/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3887 - acc: 0.8609 - val_loss: 0.5466 - val_acc: 0.8191\n",
            "Epoch 69/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3823 - acc: 0.8638 - val_loss: 0.5430 - val_acc: 0.8201\n",
            "Epoch 70/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3799 - acc: 0.8637 - val_loss: 0.5542 - val_acc: 0.8171\n",
            "Epoch 71/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3764 - acc: 0.8649 - val_loss: 0.5415 - val_acc: 0.8210\n",
            "Epoch 72/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3733 - acc: 0.8637 - val_loss: 0.5524 - val_acc: 0.8203\n",
            "Epoch 73/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3743 - acc: 0.8670 - val_loss: 0.5577 - val_acc: 0.8150\n",
            "Epoch 74/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3673 - acc: 0.8672 - val_loss: 0.5532 - val_acc: 0.8197\n",
            "Epoch 75/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3650 - acc: 0.8686 - val_loss: 0.5488 - val_acc: 0.8182\n",
            "Epoch 76/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3618 - acc: 0.8716 - val_loss: 0.5517 - val_acc: 0.8197\n",
            "Epoch 77/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3619 - acc: 0.8696 - val_loss: 0.5531 - val_acc: 0.8170\n",
            "Epoch 78/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3543 - acc: 0.8724 - val_loss: 0.5609 - val_acc: 0.8172\n",
            "Epoch 79/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3483 - acc: 0.8751 - val_loss: 0.5537 - val_acc: 0.8201\n",
            "Epoch 80/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3447 - acc: 0.8759 - val_loss: 0.5473 - val_acc: 0.8163\n",
            "Epoch 81/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3497 - acc: 0.8752 - val_loss: 0.5542 - val_acc: 0.8203\n",
            "Epoch 82/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3454 - acc: 0.8766 - val_loss: 0.5492 - val_acc: 0.8213\n",
            "Epoch 83/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3364 - acc: 0.8790 - val_loss: 0.5459 - val_acc: 0.8220\n",
            "Epoch 84/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3379 - acc: 0.8791 - val_loss: 0.5478 - val_acc: 0.8230\n",
            "Epoch 85/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3326 - acc: 0.8810 - val_loss: 0.5449 - val_acc: 0.8248\n",
            "Epoch 86/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3330 - acc: 0.8796 - val_loss: 0.5617 - val_acc: 0.8171\n",
            "Epoch 87/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3284 - acc: 0.8819 - val_loss: 0.5566 - val_acc: 0.8203\n",
            "Epoch 88/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3260 - acc: 0.8820 - val_loss: 0.5436 - val_acc: 0.8289\n",
            "Epoch 89/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3249 - acc: 0.8823 - val_loss: 0.5513 - val_acc: 0.8215\n",
            "Epoch 90/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3210 - acc: 0.8845 - val_loss: 0.5696 - val_acc: 0.8153\n",
            "Epoch 91/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3211 - acc: 0.8842 - val_loss: 0.5594 - val_acc: 0.8242\n",
            "Epoch 92/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3141 - acc: 0.8856 - val_loss: 0.5627 - val_acc: 0.8246\n",
            "Epoch 93/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3095 - acc: 0.8883 - val_loss: 0.5533 - val_acc: 0.8243\n",
            "Epoch 94/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3121 - acc: 0.8872 - val_loss: 0.5541 - val_acc: 0.8209\n",
            "Epoch 95/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.3109 - acc: 0.8881 - val_loss: 0.5515 - val_acc: 0.8226\n",
            "Epoch 96/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3110 - acc: 0.8869 - val_loss: 0.5616 - val_acc: 0.8235\n",
            "Epoch 97/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.3078 - acc: 0.8902 - val_loss: 0.5435 - val_acc: 0.8270\n",
            "Epoch 98/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.2981 - acc: 0.8922 - val_loss: 0.5615 - val_acc: 0.8226\n",
            "Epoch 99/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.2983 - acc: 0.8928 - val_loss: 0.5547 - val_acc: 0.8237\n",
            "Epoch 100/100\n",
            "40000/40000 [==============================] - 19s 472us/step - loss: 0.2948 - acc: 0.8935 - val_loss: 0.5574 - val_acc: 0.8249\n",
            "Test loss: 0.5574424630641938\n",
            "Test accuracy: 0.8249\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "validation loss: 0.5574424630641938\n",
            "validation accuracy: 0.8249\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "pr2QdR6RVLAQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 4381
        },
        "outputId": "e3c155dd-8fc9-4984-fea7-b59b3b383770"
      },
      "cell_type": "code",
      "source": [
        "#get the new training and validation set on fold Number 5 by calling kfold_validation\n",
        "model = kfold_validation(5,x_folds,y_folds)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "number of samples in training set:  40000\n",
            "number of samples in validation set:  10000\n",
            "Shape of images:  (32, 32, 3)\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 469us/step - loss: 1.9510 - acc: 0.2587\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 439us/step - loss: 1.6944 - acc: 0.3673\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 433us/step - loss: 1.5790 - acc: 0.4127\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 444us/step - loss: 1.4899 - acc: 0.4474\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 1.4288 - acc: 0.4773\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 437us/step - loss: 1.3664 - acc: 0.5032\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 438us/step - loss: 1.3228 - acc: 0.5236\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 444us/step - loss: 1.2780 - acc: 0.5392\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 17s 431us/step - loss: 1.2436 - acc: 0.5543\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 491us/step - loss: 1.2053 - acc: 0.5669\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 20s 496us/step - loss: 1.0977 - acc: 0.6078 - val_loss: 1.0020 - val_acc: 0.6429\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 1.0643 - acc: 0.6201 - val_loss: 0.9838 - val_acc: 0.6542\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 1.0310 - acc: 0.6310 - val_loss: 0.9546 - val_acc: 0.6595\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 1.0019 - acc: 0.6445 - val_loss: 0.9054 - val_acc: 0.6808\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.9824 - acc: 0.6497 - val_loss: 0.8972 - val_acc: 0.6819\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.9552 - acc: 0.6612 - val_loss: 0.8593 - val_acc: 0.6983\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.9257 - acc: 0.6740 - val_loss: 0.8380 - val_acc: 0.7071\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.9123 - acc: 0.6772 - val_loss: 0.8189 - val_acc: 0.7156\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.8856 - acc: 0.6855 - val_loss: 0.8353 - val_acc: 0.7034\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.8631 - acc: 0.6917 - val_loss: 0.7819 - val_acc: 0.7274\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.8483 - acc: 0.6988 - val_loss: 0.7751 - val_acc: 0.7300\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.8248 - acc: 0.7083 - val_loss: 0.7584 - val_acc: 0.7349\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.8107 - acc: 0.7144 - val_loss: 0.7342 - val_acc: 0.7472\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.7974 - acc: 0.7189 - val_loss: 0.7616 - val_acc: 0.7309\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.7827 - acc: 0.7215 - val_loss: 0.7176 - val_acc: 0.7494\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.7660 - acc: 0.7290 - val_loss: 0.7418 - val_acc: 0.7433\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.7515 - acc: 0.7361 - val_loss: 0.7107 - val_acc: 0.7522\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.7346 - acc: 0.7398 - val_loss: 0.7097 - val_acc: 0.7536\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.7258 - acc: 0.7436 - val_loss: 0.6772 - val_acc: 0.7664\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.7156 - acc: 0.7485 - val_loss: 0.6838 - val_acc: 0.7607\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.6999 - acc: 0.7518 - val_loss: 0.6561 - val_acc: 0.7687\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.6897 - acc: 0.7558 - val_loss: 0.6617 - val_acc: 0.7660\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.6785 - acc: 0.7612 - val_loss: 0.6630 - val_acc: 0.7683\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.6707 - acc: 0.7629 - val_loss: 0.6666 - val_acc: 0.7668\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.6581 - acc: 0.7704 - val_loss: 0.6470 - val_acc: 0.7728\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.6529 - acc: 0.7711 - val_loss: 0.6321 - val_acc: 0.7791\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.6413 - acc: 0.7732 - val_loss: 0.6457 - val_acc: 0.7759\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.6357 - acc: 0.7746 - val_loss: 0.6204 - val_acc: 0.7851\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.6271 - acc: 0.7777 - val_loss: 0.6273 - val_acc: 0.7812\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.6190 - acc: 0.7822 - val_loss: 0.6194 - val_acc: 0.7844\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.6119 - acc: 0.7839 - val_loss: 0.6082 - val_acc: 0.7850\n",
            "Epoch 32/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.6015 - acc: 0.7879 - val_loss: 0.6054 - val_acc: 0.7877\n",
            "Epoch 33/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5950 - acc: 0.7898 - val_loss: 0.5946 - val_acc: 0.7930\n",
            "Epoch 34/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5872 - acc: 0.7925 - val_loss: 0.5968 - val_acc: 0.7918\n",
            "Epoch 35/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5821 - acc: 0.7935 - val_loss: 0.5879 - val_acc: 0.7940\n",
            "Epoch 36/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5727 - acc: 0.7943 - val_loss: 0.5878 - val_acc: 0.7978\n",
            "Epoch 37/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5642 - acc: 0.8008 - val_loss: 0.6008 - val_acc: 0.7939\n",
            "Epoch 38/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5554 - acc: 0.8010 - val_loss: 0.5770 - val_acc: 0.8006\n",
            "Epoch 39/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5513 - acc: 0.8030 - val_loss: 0.5629 - val_acc: 0.8029\n",
            "Epoch 40/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5433 - acc: 0.8083 - val_loss: 0.5722 - val_acc: 0.7994\n",
            "Epoch 41/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5336 - acc: 0.8113 - val_loss: 0.5711 - val_acc: 0.8036\n",
            "Epoch 42/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.5312 - acc: 0.8117 - val_loss: 0.5691 - val_acc: 0.8025\n",
            "Epoch 43/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5255 - acc: 0.8140 - val_loss: 0.5644 - val_acc: 0.8026\n",
            "Epoch 44/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.5154 - acc: 0.8169 - val_loss: 0.5652 - val_acc: 0.8028\n",
            "Epoch 45/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5137 - acc: 0.8177 - val_loss: 0.5623 - val_acc: 0.8064\n",
            "Epoch 46/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.5085 - acc: 0.8199 - val_loss: 0.5516 - val_acc: 0.8080\n",
            "Epoch 47/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.5049 - acc: 0.8221 - val_loss: 0.5628 - val_acc: 0.8083\n",
            "Epoch 48/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4995 - acc: 0.8230 - val_loss: 0.5765 - val_acc: 0.8004\n",
            "Epoch 49/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4945 - acc: 0.8242 - val_loss: 0.5711 - val_acc: 0.8017\n",
            "Epoch 50/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4862 - acc: 0.8260 - val_loss: 0.5488 - val_acc: 0.8098\n",
            "Epoch 51/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.4862 - acc: 0.8269 - val_loss: 0.5469 - val_acc: 0.8124\n",
            "Epoch 52/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.4764 - acc: 0.8312 - val_loss: 0.5370 - val_acc: 0.8146\n",
            "Epoch 53/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4702 - acc: 0.8324 - val_loss: 0.5407 - val_acc: 0.8141\n",
            "Epoch 54/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4639 - acc: 0.8332 - val_loss: 0.5495 - val_acc: 0.8145\n",
            "Epoch 55/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4631 - acc: 0.8347 - val_loss: 0.5387 - val_acc: 0.8151\n",
            "Epoch 56/100\n",
            "40000/40000 [==============================] - 19s 473us/step - loss: 0.4549 - acc: 0.8368 - val_loss: 0.5337 - val_acc: 0.8139\n",
            "Epoch 57/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4507 - acc: 0.8400 - val_loss: 0.5334 - val_acc: 0.8162\n",
            "Epoch 58/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4445 - acc: 0.8420 - val_loss: 0.5425 - val_acc: 0.8169\n",
            "Epoch 59/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4410 - acc: 0.8418 - val_loss: 0.5306 - val_acc: 0.8163\n",
            "Epoch 60/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.4392 - acc: 0.8425 - val_loss: 0.5380 - val_acc: 0.8162\n",
            "Epoch 61/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.4306 - acc: 0.8454 - val_loss: 0.5508 - val_acc: 0.8149\n",
            "Epoch 62/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.4331 - acc: 0.8426 - val_loss: 0.5341 - val_acc: 0.8186\n",
            "Epoch 63/100\n",
            "40000/40000 [==============================] - 19s 474us/step - loss: 0.4216 - acc: 0.8503 - val_loss: 0.5310 - val_acc: 0.8180\n",
            "Epoch 64/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.4274 - acc: 0.8467 - val_loss: 0.5314 - val_acc: 0.8198\n",
            "Epoch 65/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.4213 - acc: 0.8481 - val_loss: 0.5306 - val_acc: 0.8193\n",
            "Epoch 66/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.4151 - acc: 0.8515 - val_loss: 0.5254 - val_acc: 0.8218\n",
            "Epoch 67/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.4088 - acc: 0.8526 - val_loss: 0.5345 - val_acc: 0.8219\n",
            "Epoch 68/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.4057 - acc: 0.8535 - val_loss: 0.5514 - val_acc: 0.8173\n",
            "Epoch 69/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.4029 - acc: 0.8557 - val_loss: 0.5259 - val_acc: 0.8217\n",
            "Epoch 70/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3981 - acc: 0.8567 - val_loss: 0.5258 - val_acc: 0.8243\n",
            "Epoch 71/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.3943 - acc: 0.8596 - val_loss: 0.5344 - val_acc: 0.8200\n",
            "Epoch 72/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3918 - acc: 0.8616 - val_loss: 0.5363 - val_acc: 0.8223\n",
            "Epoch 73/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3853 - acc: 0.8612 - val_loss: 0.5379 - val_acc: 0.8192\n",
            "Epoch 74/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3881 - acc: 0.8594 - val_loss: 0.5436 - val_acc: 0.8175\n",
            "Epoch 75/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3788 - acc: 0.8644 - val_loss: 0.5239 - val_acc: 0.8262\n",
            "Epoch 76/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3848 - acc: 0.8609 - val_loss: 0.5331 - val_acc: 0.8266\n",
            "Epoch 77/100\n",
            "40000/40000 [==============================] - 19s 476us/step - loss: 0.3703 - acc: 0.8669 - val_loss: 0.5200 - val_acc: 0.8257\n",
            "Epoch 78/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3728 - acc: 0.8656 - val_loss: 0.5208 - val_acc: 0.8240\n",
            "Epoch 79/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.3685 - acc: 0.8684 - val_loss: 0.5350 - val_acc: 0.8216\n",
            "Epoch 80/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3598 - acc: 0.8700 - val_loss: 0.5146 - val_acc: 0.8311\n",
            "Epoch 81/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3587 - acc: 0.8709 - val_loss: 0.5127 - val_acc: 0.8308\n",
            "Epoch 82/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3602 - acc: 0.8700 - val_loss: 0.5131 - val_acc: 0.8265\n",
            "Epoch 83/100\n",
            "40000/40000 [==============================] - 19s 475us/step - loss: 0.3578 - acc: 0.8722 - val_loss: 0.5247 - val_acc: 0.8258\n",
            "Epoch 84/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3553 - acc: 0.8747 - val_loss: 0.5068 - val_acc: 0.8314\n",
            "Epoch 85/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3493 - acc: 0.8758 - val_loss: 0.5078 - val_acc: 0.8313\n",
            "Epoch 86/100\n",
            "40000/40000 [==============================] - 19s 482us/step - loss: 0.3495 - acc: 0.8756 - val_loss: 0.5109 - val_acc: 0.8318\n",
            "Epoch 87/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3427 - acc: 0.8769 - val_loss: 0.5058 - val_acc: 0.8294\n",
            "Epoch 88/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3406 - acc: 0.8772 - val_loss: 0.5222 - val_acc: 0.8280\n",
            "Epoch 89/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3427 - acc: 0.8761 - val_loss: 0.5233 - val_acc: 0.8243\n",
            "Epoch 90/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3331 - acc: 0.8790 - val_loss: 0.5100 - val_acc: 0.8324\n",
            "Epoch 91/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3333 - acc: 0.8794 - val_loss: 0.5114 - val_acc: 0.8306\n",
            "Epoch 92/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3319 - acc: 0.8807 - val_loss: 0.5218 - val_acc: 0.8258\n",
            "Epoch 93/100\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 0.3291 - acc: 0.8814 - val_loss: 0.5374 - val_acc: 0.8263\n",
            "Epoch 94/100\n",
            "40000/40000 [==============================] - 19s 478us/step - loss: 0.3274 - acc: 0.8830 - val_loss: 0.5177 - val_acc: 0.8319\n",
            "Epoch 95/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3257 - acc: 0.8844 - val_loss: 0.5177 - val_acc: 0.8309\n",
            "Epoch 96/100\n",
            "40000/40000 [==============================] - 19s 481us/step - loss: 0.3205 - acc: 0.8854 - val_loss: 0.5109 - val_acc: 0.8302\n",
            "Epoch 97/100\n",
            "40000/40000 [==============================] - 19s 480us/step - loss: 0.3169 - acc: 0.8856 - val_loss: 0.5176 - val_acc: 0.8318\n",
            "Epoch 98/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3156 - acc: 0.8873 - val_loss: 0.5268 - val_acc: 0.8307\n",
            "Epoch 99/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3152 - acc: 0.8882 - val_loss: 0.5201 - val_acc: 0.8286\n",
            "Epoch 100/100\n",
            "40000/40000 [==============================] - 19s 479us/step - loss: 0.3137 - acc: 0.8861 - val_loss: 0.5325 - val_acc: 0.8260\n",
            "Test loss: 0.53251511759758\n",
            "Test accuracy: 0.826\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "validation loss: 0.53251511759758\n",
            "validation accuracy: 0.826\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "LJC_8SKdaBfq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "6230422c-a76c-4cce-c36e-40790077b1fb"
      },
      "cell_type": "code",
      "source": [
        "mean_accuracy = np.mean(validation_accuracy)\n",
        "print('Mean accuracy of K-fold validation ',mean_accuracy)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mean accuracy of K-fold validation  0.8278800000000001\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "SHUvu-B-VcNi",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "cfc5ebbd-536d-4b70-cea6-e10ae818eb80"
      },
      "cell_type": "code",
      "source": [
        "#evaluating the model on test set\n",
        "test_model(model,x_test,y_test)"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Test loss: 0.5554813200831413\n",
            "Test accuracy: 0.8206\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}