hw2_model2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "hw2_model2.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_model2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "ZmBJHRH-YqMx",
        "colab_type": "code",
        "outputId": "e2a30d02-c0fc-44fb-e5ee-29aa208b79a7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "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": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "uURXeJQPdJk6",
        "colab_type": "code",
        "outputId": "fe30760d-ecc2-43db-b14e-81089e0ab1b7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        }
      },
      "cell_type": "code",
      "source": [
        "# Required variables are declared\n",
        "batch_size = 32\n",
        "num_classes = 10\n",
        "epochs = 100\n",
        "\n",
        "#Loading and shuffling Cifar10 dataset\n",
        "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
        "x_train, y_train = shuffle(x_train, y_train, random_state=0)\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\n",
        "\n",
        "# Dividing data into training and validation sets\n",
        "train_x = x_train[:split_index]\n",
        "train_y = y_train[:split_index]\n",
        "\n",
        "validate_x = x_train[split_index:]\n",
        "validate_y = y_train[split_index:]\n",
        "\n",
        "print(\"\\nnumber of samples in training set: \",len(train_x))\n",
        "print(\"number of samples in validation set: \",len(validate_x))\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "print(\"Shape of images: \",input_shape)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
            "170500096/170498071 [==============================] - 95s 1us/step\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"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "4CZrpmFHdTTa",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Defined the model\n",
        "#def getModel(input_shape):\n",
        "  \n",
        "\t#model = Sequential()\n",
        "  #model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape))\n",
        "  #model.add(Conv2D(32, kernel_size=(3, 3), activation='relu'))\n",
        "\t#model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "  #model.add(Dropout(0.25))\n",
        "\n",
        "\t#model.add(Conv2D(64, kernel_size=(3, 3), activation='relu', padding='same'))\n",
        "  #model.add(Conv2D(64, kernel_size=(3, 3), activation='relu'))\n",
        "  #model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "  #model.add(Dropout(0.25))\n",
        "\n",
        "\t#model.add(Flatten())\n",
        "  #model.add(Dense(512, activation='relu'))\n",
        "  #model.add(Dropout(0.5))\n",
        "  #model.add(Dense(num_classes, activation='softmax'))\n",
        "\n",
        "\t#model.summary()\n",
        "\t#model.compile(loss=keras.losses.categorical_crossentropy,optimizer=Adam(lr=0.0001, decay=1e-6),metrics=['accuracy'])\n",
        "\t\n",
        "  #return model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "9tBf_9CRFyx8",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#Both the commented and executed model gave the same accuracy with slight difference.\n",
        "def getModel(input_shape):\n",
        "  model = Sequential()\n",
        "  model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape))\n",
        "  model.add(Conv2D(64, kernel_size=(3, 3), activation='relu'))\n",
        "  model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "  model.add(Dropout(0.25))\n",
        "\n",
        "  model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))\n",
        "  model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "  model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))\n",
        "  model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "  model.add(Dropout(0.25))\n",
        "  \n",
        "  model.add(Flatten())\n",
        "  model.add(Dense(1024, activation='relu'))\n",
        "  model.add(Dropout(0.5))\n",
        "\n",
        "  model.add(Dense(10, activation='softmax'))\n",
        "  model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=Adam(lr=0.0001, decay=1e-6),\n",
        "              metrics=['accuracy'])\n",
        "  return model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "H78KdQ-sgtFY",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def train_model(model):\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": "2cmnrQZwguAU",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Model is tested in this method\n",
        "def test_model():\n",
        "\t\n",
        "\tscore = model.evaluate(validate_x, validate_y, verbose=0)\n",
        "\t\n",
        "\tprint('\\n\\nTest loss:', score[0])\n",
        "\tprint('Test accuracy:', score[1])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "CM3Ya4qcxi7J",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#Plotted the accuracy against number of epochs\n",
        "def plot_training_graph(training_history):\n",
        "\t\n",
        "\tplotaccuracy = plt.plot(range(1,epochs+1), training_history.history['acc'],range(1,epochs+1), training_history.history['val_acc'])\n",
        "\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": "8kRU0v8jyz6H",
        "colab_type": "code",
        "outputId": "6bbf7052-754e-4159-d570-9d2055880204",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 4507
        }
      },
      "cell_type": "code",
      "source": [
        "# Calling all the models\n",
        "model = getModel(input_shape)\n",
        "History = train_model(model)\n",
        "test_model()\n",
        "\n",
        "plot_training_graph(History)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 24s 597us/step - loss: 1.9013 - acc: 0.2878\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 20s 492us/step - loss: 1.5863 - acc: 0.4129\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 23s 585us/step - loss: 1.4565 - acc: 0.4667\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 468us/step - loss: 1.3616 - acc: 0.5058\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 23s 581us/step - loss: 1.2870 - acc: 0.5363\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 471us/step - loss: 1.2348 - acc: 0.5585\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 18s 457us/step - loss: 1.1842 - acc: 0.5762\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 23s 582us/step - loss: 1.1487 - acc: 0.5933\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 24s 589us/step - loss: 1.1062 - acc: 0.6085\n",
            "Epoch 1/1\n",
            "40000/40000 [==============================] - 19s 477us/step - loss: 1.0700 - acc: 0.6227\n",
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/100\n",
            "40000/40000 [==============================] - 20s 505us/step - loss: 0.9511 - acc: 0.6666 - val_loss: 0.8916 - val_acc: 0.6944\n",
            "Epoch 2/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.9130 - acc: 0.6784 - val_loss: 0.8422 - val_acc: 0.7067\n",
            "Epoch 3/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.8745 - acc: 0.6943 - val_loss: 0.8162 - val_acc: 0.7163\n",
            "Epoch 4/100\n",
            "40000/40000 [==============================] - 20s 488us/step - loss: 0.8440 - acc: 0.7029 - val_loss: 0.8077 - val_acc: 0.7193\n",
            "Epoch 5/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.8185 - acc: 0.7124 - val_loss: 0.7647 - val_acc: 0.7322\n",
            "Epoch 6/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.7857 - acc: 0.7253 - val_loss: 0.7570 - val_acc: 0.7329\n",
            "Epoch 7/100\n",
            "40000/40000 [==============================] - 20s 491us/step - loss: 0.7664 - acc: 0.7329 - val_loss: 0.7417 - val_acc: 0.7386\n",
            "Epoch 8/100\n",
            "40000/40000 [==============================] - 20s 500us/step - loss: 0.7397 - acc: 0.7394 - val_loss: 0.7276 - val_acc: 0.7450\n",
            "Epoch 9/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.7187 - acc: 0.7481 - val_loss: 0.7181 - val_acc: 0.7478\n",
            "Epoch 10/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.6937 - acc: 0.7572 - val_loss: 0.7130 - val_acc: 0.7477\n",
            "Epoch 11/100\n",
            "40000/40000 [==============================] - 21s 516us/step - loss: 0.6777 - acc: 0.7632 - val_loss: 0.6990 - val_acc: 0.7527\n",
            "Epoch 12/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.6559 - acc: 0.7699 - val_loss: 0.6918 - val_acc: 0.7566\n",
            "Epoch 13/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.6421 - acc: 0.7743 - val_loss: 0.6748 - val_acc: 0.7654\n",
            "Epoch 14/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.6196 - acc: 0.7831 - val_loss: 0.6716 - val_acc: 0.7667\n",
            "Epoch 15/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.6053 - acc: 0.7879 - val_loss: 0.6626 - val_acc: 0.7672\n",
            "Epoch 16/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.5902 - acc: 0.7928 - val_loss: 0.6573 - val_acc: 0.7689\n",
            "Epoch 17/100\n",
            "40000/40000 [==============================] - 20s 512us/step - loss: 0.5714 - acc: 0.8005 - val_loss: 0.6612 - val_acc: 0.7695\n",
            "Epoch 18/100\n",
            "40000/40000 [==============================] - 21s 513us/step - loss: 0.5607 - acc: 0.8026 - val_loss: 0.6444 - val_acc: 0.7734\n",
            "Epoch 19/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.5476 - acc: 0.8079 - val_loss: 0.6416 - val_acc: 0.7801\n",
            "Epoch 20/100\n",
            "40000/40000 [==============================] - 21s 516us/step - loss: 0.5324 - acc: 0.8109 - val_loss: 0.6338 - val_acc: 0.7818\n",
            "Epoch 21/100\n",
            "40000/40000 [==============================] - 20s 492us/step - loss: 0.5196 - acc: 0.8159 - val_loss: 0.6335 - val_acc: 0.7827\n",
            "Epoch 22/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.5078 - acc: 0.8221 - val_loss: 0.6263 - val_acc: 0.7858\n",
            "Epoch 23/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.4943 - acc: 0.8262 - val_loss: 0.6215 - val_acc: 0.7884\n",
            "Epoch 24/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.4840 - acc: 0.8289 - val_loss: 0.6392 - val_acc: 0.7826\n",
            "Epoch 25/100\n",
            "40000/40000 [==============================] - 21s 521us/step - loss: 0.4697 - acc: 0.8353 - val_loss: 0.6243 - val_acc: 0.7857\n",
            "Epoch 26/100\n",
            "40000/40000 [==============================] - 21s 521us/step - loss: 0.4624 - acc: 0.8364 - val_loss: 0.6320 - val_acc: 0.7841\n",
            "Epoch 27/100\n",
            "40000/40000 [==============================] - 21s 521us/step - loss: 0.4469 - acc: 0.8425 - val_loss: 0.6206 - val_acc: 0.7899\n",
            "Epoch 28/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.4370 - acc: 0.8462 - val_loss: 0.6227 - val_acc: 0.7889\n",
            "Epoch 29/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.4254 - acc: 0.8492 - val_loss: 0.6107 - val_acc: 0.7917\n",
            "Epoch 30/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.4164 - acc: 0.8531 - val_loss: 0.6099 - val_acc: 0.7947\n",
            "Epoch 31/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.4083 - acc: 0.8554 - val_loss: 0.6239 - val_acc: 0.7944\n",
            "Epoch 32/100\n",
            "40000/40000 [==============================] - 21s 518us/step - loss: 0.3979 - acc: 0.8585 - val_loss: 0.6123 - val_acc: 0.7936\n",
            "Epoch 33/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.3900 - acc: 0.8622 - val_loss: 0.6084 - val_acc: 0.7975\n",
            "Epoch 34/100\n",
            "40000/40000 [==============================] - 21s 518us/step - loss: 0.3808 - acc: 0.8657 - val_loss: 0.6234 - val_acc: 0.7961\n",
            "Epoch 35/100\n",
            "40000/40000 [==============================] - 21s 524us/step - loss: 0.3683 - acc: 0.8691 - val_loss: 0.6191 - val_acc: 0.7947\n",
            "Epoch 36/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.3608 - acc: 0.8719 - val_loss: 0.6127 - val_acc: 0.7975\n",
            "Epoch 37/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.3543 - acc: 0.8739 - val_loss: 0.6251 - val_acc: 0.7944\n",
            "Epoch 38/100\n",
            "40000/40000 [==============================] - 21s 523us/step - loss: 0.3446 - acc: 0.8778 - val_loss: 0.6163 - val_acc: 0.8001\n",
            "Epoch 39/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.3320 - acc: 0.8821 - val_loss: 0.6340 - val_acc: 0.7973\n",
            "Epoch 40/100\n",
            "40000/40000 [==============================] - 21s 517us/step - loss: 0.3303 - acc: 0.8808 - val_loss: 0.6344 - val_acc: 0.7998\n",
            "Epoch 41/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.3241 - acc: 0.8834 - val_loss: 0.6220 - val_acc: 0.7968\n",
            "Epoch 42/100\n",
            "40000/40000 [==============================] - 21s 521us/step - loss: 0.3109 - acc: 0.8885 - val_loss: 0.6134 - val_acc: 0.8018\n",
            "Epoch 43/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.3086 - acc: 0.8895 - val_loss: 0.6210 - val_acc: 0.8000\n",
            "Epoch 44/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.2979 - acc: 0.8940 - val_loss: 0.6410 - val_acc: 0.7987\n",
            "Epoch 45/100\n",
            "40000/40000 [==============================] - 21s 518us/step - loss: 0.2906 - acc: 0.8956 - val_loss: 0.6794 - val_acc: 0.7930\n",
            "Epoch 46/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.2851 - acc: 0.8978 - val_loss: 0.6259 - val_acc: 0.8017\n",
            "Epoch 47/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.2839 - acc: 0.8978 - val_loss: 0.6279 - val_acc: 0.8009\n",
            "Epoch 48/100\n",
            "40000/40000 [==============================] - 21s 517us/step - loss: 0.2748 - acc: 0.9017 - val_loss: 0.6351 - val_acc: 0.8045\n",
            "Epoch 49/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.2648 - acc: 0.9051 - val_loss: 0.6420 - val_acc: 0.7994\n",
            "Epoch 50/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.2603 - acc: 0.9065 - val_loss: 0.6563 - val_acc: 0.8010\n",
            "Epoch 51/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.2529 - acc: 0.9093 - val_loss: 0.6509 - val_acc: 0.8006\n",
            "Epoch 52/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.2534 - acc: 0.9092 - val_loss: 0.6632 - val_acc: 0.7957\n",
            "Epoch 53/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.2506 - acc: 0.9119 - val_loss: 0.6504 - val_acc: 0.8017\n",
            "Epoch 54/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.2412 - acc: 0.9134 - val_loss: 0.6548 - val_acc: 0.8020\n",
            "Epoch 55/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.2342 - acc: 0.9154 - val_loss: 0.6696 - val_acc: 0.7989\n",
            "Epoch 56/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.2305 - acc: 0.9179 - val_loss: 0.6653 - val_acc: 0.8034\n",
            "Epoch 57/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.2226 - acc: 0.9190 - val_loss: 0.6728 - val_acc: 0.8014\n",
            "Epoch 58/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.2235 - acc: 0.9188 - val_loss: 0.6770 - val_acc: 0.8019\n",
            "Epoch 59/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.2153 - acc: 0.9234 - val_loss: 0.6695 - val_acc: 0.7998\n",
            "Epoch 60/100\n",
            "40000/40000 [==============================] - 21s 516us/step - loss: 0.2129 - acc: 0.9237 - val_loss: 0.6691 - val_acc: 0.8057\n",
            "Epoch 61/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.2135 - acc: 0.9232 - val_loss: 0.6662 - val_acc: 0.8049\n",
            "Epoch 62/100\n",
            "40000/40000 [==============================] - 20s 493us/step - loss: 0.2074 - acc: 0.9244 - val_loss: 0.6772 - val_acc: 0.8031\n",
            "Epoch 63/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.2011 - acc: 0.9290 - val_loss: 0.6899 - val_acc: 0.7998\n",
            "Epoch 64/100\n",
            "40000/40000 [==============================] - 20s 490us/step - loss: 0.1983 - acc: 0.9297 - val_loss: 0.6903 - val_acc: 0.8019\n",
            "Epoch 65/100\n",
            "40000/40000 [==============================] - 20s 489us/step - loss: 0.1946 - acc: 0.9295 - val_loss: 0.6880 - val_acc: 0.8038\n",
            "Epoch 66/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.1904 - acc: 0.9317 - val_loss: 0.6836 - val_acc: 0.8020\n",
            "Epoch 67/100\n",
            "40000/40000 [==============================] - 21s 521us/step - loss: 0.1836 - acc: 0.9341 - val_loss: 0.7033 - val_acc: 0.8050\n",
            "Epoch 68/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.1896 - acc: 0.9322 - val_loss: 0.7034 - val_acc: 0.7978\n",
            "Epoch 69/100\n",
            "40000/40000 [==============================] - 21s 518us/step - loss: 0.1811 - acc: 0.9344 - val_loss: 0.7167 - val_acc: 0.8065\n",
            "Epoch 70/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.1752 - acc: 0.9374 - val_loss: 0.7051 - val_acc: 0.8056\n",
            "Epoch 71/100\n",
            "40000/40000 [==============================] - 21s 521us/step - loss: 0.1768 - acc: 0.9365 - val_loss: 0.7066 - val_acc: 0.8087\n",
            "Epoch 72/100\n",
            "40000/40000 [==============================] - 21s 519us/step - loss: 0.1706 - acc: 0.9385 - val_loss: 0.7132 - val_acc: 0.8035\n",
            "Epoch 73/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.1670 - acc: 0.9410 - val_loss: 0.7114 - val_acc: 0.8066\n",
            "Epoch 74/100\n",
            "40000/40000 [==============================] - 21s 516us/step - loss: 0.1704 - acc: 0.9397 - val_loss: 0.7046 - val_acc: 0.8042\n",
            "Epoch 75/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.1602 - acc: 0.9424 - val_loss: 0.7232 - val_acc: 0.8040\n",
            "Epoch 76/100\n",
            "40000/40000 [==============================] - 21s 517us/step - loss: 0.1667 - acc: 0.9415 - val_loss: 0.6940 - val_acc: 0.8078\n",
            "Epoch 77/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.1605 - acc: 0.9434 - val_loss: 0.7282 - val_acc: 0.8032\n",
            "Epoch 78/100\n",
            "40000/40000 [==============================] - 21s 531us/step - loss: 0.1539 - acc: 0.9463 - val_loss: 0.7119 - val_acc: 0.8072\n",
            "Epoch 79/100\n",
            "40000/40000 [==============================] - 20s 499us/step - loss: 0.1544 - acc: 0.9444 - val_loss: 0.7315 - val_acc: 0.8048\n",
            "Epoch 80/100\n",
            "40000/40000 [==============================] - 20s 492us/step - loss: 0.1550 - acc: 0.9458 - val_loss: 0.7116 - val_acc: 0.8056\n",
            "Epoch 81/100\n",
            "40000/40000 [==============================] - 20s 512us/step - loss: 0.1493 - acc: 0.9465 - val_loss: 0.7333 - val_acc: 0.8031\n",
            "Epoch 82/100\n",
            "40000/40000 [==============================] - 20s 511us/step - loss: 0.1509 - acc: 0.9484 - val_loss: 0.7214 - val_acc: 0.8063\n",
            "Epoch 83/100\n",
            "40000/40000 [==============================] - 21s 517us/step - loss: 0.1482 - acc: 0.9467 - val_loss: 0.7239 - val_acc: 0.8062\n",
            "Epoch 84/100\n",
            "40000/40000 [==============================] - 21s 520us/step - loss: 0.1424 - acc: 0.9496 - val_loss: 0.7291 - val_acc: 0.8079\n",
            "Epoch 85/100\n",
            "40000/40000 [==============================] - 21s 523us/step - loss: 0.1444 - acc: 0.9490 - val_loss: 0.7555 - val_acc: 0.8058\n",
            "Epoch 86/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.1422 - acc: 0.9491 - val_loss: 0.7274 - val_acc: 0.8056\n",
            "Epoch 87/100\n",
            "40000/40000 [==============================] - 21s 518us/step - loss: 0.1354 - acc: 0.9514 - val_loss: 0.7480 - val_acc: 0.8037\n",
            "Epoch 88/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.1343 - acc: 0.9526 - val_loss: 0.7487 - val_acc: 0.8052\n",
            "Epoch 89/100\n",
            "40000/40000 [==============================] - 21s 516us/step - loss: 0.1347 - acc: 0.9535 - val_loss: 0.7671 - val_acc: 0.8044\n",
            "Epoch 90/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.1346 - acc: 0.9516 - val_loss: 0.7473 - val_acc: 0.8075\n",
            "Epoch 91/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.1351 - acc: 0.9519 - val_loss: 0.7389 - val_acc: 0.8060\n",
            "Epoch 92/100\n",
            "40000/40000 [==============================] - 21s 517us/step - loss: 0.1308 - acc: 0.9544 - val_loss: 0.7456 - val_acc: 0.8104\n",
            "Epoch 93/100\n",
            "40000/40000 [==============================] - 21s 522us/step - loss: 0.1311 - acc: 0.9530 - val_loss: 0.7625 - val_acc: 0.8080\n",
            "Epoch 94/100\n",
            "40000/40000 [==============================] - 21s 518us/step - loss: 0.1258 - acc: 0.9564 - val_loss: 0.7607 - val_acc: 0.8085\n",
            "Epoch 95/100\n",
            "40000/40000 [==============================] - 21s 515us/step - loss: 0.1238 - acc: 0.9568 - val_loss: 0.7794 - val_acc: 0.8017\n",
            "Epoch 96/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.1255 - acc: 0.9552 - val_loss: 0.7412 - val_acc: 0.8105\n",
            "Epoch 97/100\n",
            "40000/40000 [==============================] - 19s 486us/step - loss: 0.1199 - acc: 0.9574 - val_loss: 0.7620 - val_acc: 0.8074\n",
            "Epoch 98/100\n",
            "40000/40000 [==============================] - 20s 509us/step - loss: 0.1168 - acc: 0.9593 - val_loss: 0.7883 - val_acc: 0.8084\n",
            "Epoch 99/100\n",
            "40000/40000 [==============================] - 21s 514us/step - loss: 0.1174 - acc: 0.9593 - val_loss: 0.7732 - val_acc: 0.8114\n",
            "Epoch 100/100\n",
            "40000/40000 [==============================] - 19s 487us/step - loss: 0.1192 - acc: 0.9570 - val_loss: 0.7577 - val_acc: 0.8097\n",
            "\n",
            "\n",
            "Test loss: 0.7576512030959129\n",
            "Test accuracy: 0.8097\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFYCAYAAABKymUhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "\n",
            "validation loss: 0.7576512030959129\n",
            "validation accuracy: 0.8097\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}