ProjetoDeepLearningWithLeakyReLU.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "ProjetoDeepLearningWithLeakyReLU.ipynb",
      "provenance": [],
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/masdesouza/study_artificial_intelligence/blob/main/ProjetoDeepLearningWithLeakyReLU.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CRdyQsfbrjMo",
        "outputId": "368c9cd4-89ec-4e01-dd12-ca37a2eeac09"
      },
      "source": [
        "gpu_info = !nvidia-smi\n",
        "gpu_info = '\\n'.join(gpu_info)\n",
        "if gpu_info.find('failed') >= 0:\n",
        "  print('Select the Runtime > \"Change runtime type\" menu to enable a GPU accelerator, ')\n",
        "  print('and then re-execute this cell.')\n",
        "else:\n",
        "  print(gpu_info)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Wed Apr 28 18:12:24 2021       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 465.19.01    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   34C    P0    29W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PCA7nO39UUCW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e03e409c-dd8a-4042-d5f0-43fbf5571ad2"
      },
      "source": [
        "'''Trains a simple convnet on the MNIST dataset.\n",
        "Gets to 99.25% test accuracy after 12 epochs\n",
        "(there is still a lot of margin for parameter tuning).\n",
        "16 seconds per epoch on a GRID K520 GPU.\n",
        "'''\n",
        "\n",
        "from __future__ import print_function\n",
        "import keras\n",
        "from keras.datasets import cifar10\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout, Flatten, LeakyReLU\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras import backend as K\n",
        "from keras.optimizers import SGD, Adamax, RMSprop, Adam\n",
        "\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "import numpy as np\n",
        "\n",
        "batch_size =  128\n",
        "num_classes = 10\n",
        "epochs = 120\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "# input image dimensions\n",
        "img_rows, img_cols = 32, 32\n",
        "\n",
        "# the data, split between train and test sets\n",
        "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
        "y_train = to_categorical(y_train)\n",
        "y_test = to_categorical(y_test)\n",
        "\n",
        "#input_shape = (3,32,32)\n",
        "input_shape=(32, 32, 3)\n",
        "\n",
        "#Conversão de inteiros para real\n",
        "x_train = x_train.astype('float32')\n",
        "x_test = x_test.astype('float32')\n",
        "#Normalização para o range 0-1\n",
        "x_train /= 255\n",
        "x_test /= 255\n",
        "\n",
        "\n",
        "print('x_train shape:', x_train.shape)\n",
        "print(x_train.shape[0], 'train samples')\n",
        "print(x_test.shape[0], 'test samples')\n",
        "\n",
        "# convert class vectors to binary class matrices\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=np.rollaxis(x_train, 3, 1)\n",
        "#x_test=np.rollaxis(x_test, 3, 1)\n",
        "\n",
        "\n",
        "#Construindo o modelo\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3) , kernel_initializer='he_uniform', padding='same', input_shape=input_shape))\n",
        "model.add(Conv2D(32, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(Conv2D(128, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "model.add(Conv2D(128, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(Conv2D(512, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "model.add(Conv2D(512, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(Conv2D(256, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "model.add(Conv2D(256, (3, 3), activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform', padding='same'))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(700, activation=keras.layers.LeakyReLU(alpha=0.3), kernel_initializer='he_uniform'))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(Dense(10, activation='softmax'))\n",
        "\n",
        "\n",
        "\n",
        "#Compilando o modelo\n",
        "#opt = SGD(lr=0.001, momentum=0.9)\n",
        "#opt = Adamax(learning_rate=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-07)\n",
        "opt = Adamax(learning_rate=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-07) #beta e epsilon normalmente não se ajusta, só a taxa\n",
        "\n",
        "model.compile(optimizer=opt, loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "\n",
        "\n",
        "#Avaliando o modelo com a base de teste\n",
        "\n",
        "# fit model\n",
        "history = model.fit(x_train, y_train, epochs=epochs, batch_size=batch_size, validation_data=(x_test, y_test), verbose=1)\n",
        "\n",
        "\n",
        "# evaluate model\n",
        "#_, acc = model.evaluate(x_test, y_test, verbose=0)\n",
        "#print('> %.3f' % (acc * 100.0))\n",
        "\n",
        "\n",
        "\n",
        "score = model.evaluate(x_test, y_test, verbose=0)\n",
        "\n",
        "\n",
        "print('Test loss:', ' %.3f' %  ( score[0] ) )\n",
        "print('Test accuracy:', ' %.3f' % ( score[1]* 100.0) )"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
            "170500096/170498071 [==============================] - 11s 0us/step\n",
            "x_train shape: (50000, 32, 32, 3)\n",
            "50000 train samples\n",
            "10000 test samples\n",
            "Epoch 1/120\n",
            "391/391 [==============================] - 26s 24ms/step - loss: 2.1316 - accuracy: 0.2908 - val_loss: 1.4094 - val_accuracy: 0.4955\n",
            "Epoch 2/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 1.3018 - accuracy: 0.5277 - val_loss: 1.1607 - val_accuracy: 0.5853\n",
            "Epoch 3/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 1.0357 - accuracy: 0.6303 - val_loss: 0.9556 - val_accuracy: 0.6713\n",
            "Epoch 4/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.8966 - accuracy: 0.6846 - val_loss: 0.8251 - val_accuracy: 0.7174\n",
            "Epoch 5/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.7796 - accuracy: 0.7265 - val_loss: 0.7537 - val_accuracy: 0.7420\n",
            "Epoch 6/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.6939 - accuracy: 0.7552 - val_loss: 0.6912 - val_accuracy: 0.7651\n",
            "Epoch 7/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.6493 - accuracy: 0.7778 - val_loss: 0.6965 - val_accuracy: 0.7676\n",
            "Epoch 8/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.5841 - accuracy: 0.7977 - val_loss: 0.6634 - val_accuracy: 0.7792\n",
            "Epoch 9/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.5209 - accuracy: 0.8166 - val_loss: 0.6971 - val_accuracy: 0.7784\n",
            "Epoch 10/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.4825 - accuracy: 0.8313 - val_loss: 0.6193 - val_accuracy: 0.7999\n",
            "Epoch 11/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.4359 - accuracy: 0.8464 - val_loss: 0.7193 - val_accuracy: 0.7896\n",
            "Epoch 12/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.3911 - accuracy: 0.8606 - val_loss: 0.6642 - val_accuracy: 0.8088\n",
            "Epoch 13/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.3680 - accuracy: 0.8685 - val_loss: 0.6650 - val_accuracy: 0.8120\n",
            "Epoch 14/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.3327 - accuracy: 0.8839 - val_loss: 0.6079 - val_accuracy: 0.8170\n",
            "Epoch 15/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.3133 - accuracy: 0.8893 - val_loss: 0.6589 - val_accuracy: 0.8155\n",
            "Epoch 16/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.2821 - accuracy: 0.8996 - val_loss: 0.6603 - val_accuracy: 0.8200\n",
            "Epoch 17/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.2710 - accuracy: 0.9025 - val_loss: 0.6555 - val_accuracy: 0.8213\n",
            "Epoch 18/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.2409 - accuracy: 0.9140 - val_loss: 0.7154 - val_accuracy: 0.8145\n",
            "Epoch 19/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.2259 - accuracy: 0.9204 - val_loss: 0.6793 - val_accuracy: 0.8256\n",
            "Epoch 20/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.2222 - accuracy: 0.9215 - val_loss: 0.7053 - val_accuracy: 0.8298\n",
            "Epoch 21/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.2021 - accuracy: 0.9272 - val_loss: 0.7056 - val_accuracy: 0.8206\n",
            "Epoch 22/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1932 - accuracy: 0.9318 - val_loss: 0.6524 - val_accuracy: 0.8310\n",
            "Epoch 23/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1786 - accuracy: 0.9379 - val_loss: 0.7163 - val_accuracy: 0.8304\n",
            "Epoch 24/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1673 - accuracy: 0.9404 - val_loss: 0.7742 - val_accuracy: 0.8246\n",
            "Epoch 25/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1675 - accuracy: 0.9412 - val_loss: 0.6687 - val_accuracy: 0.8357\n",
            "Epoch 26/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1554 - accuracy: 0.9442 - val_loss: 0.7441 - val_accuracy: 0.8295\n",
            "Epoch 27/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1505 - accuracy: 0.9497 - val_loss: 0.7647 - val_accuracy: 0.8341\n",
            "Epoch 28/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1430 - accuracy: 0.9495 - val_loss: 0.7579 - val_accuracy: 0.8295\n",
            "Epoch 29/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1341 - accuracy: 0.9520 - val_loss: 0.7499 - val_accuracy: 0.8352\n",
            "Epoch 30/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1395 - accuracy: 0.9516 - val_loss: 0.7905 - val_accuracy: 0.8290\n",
            "Epoch 31/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1267 - accuracy: 0.9561 - val_loss: 0.8016 - val_accuracy: 0.8357\n",
            "Epoch 32/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1250 - accuracy: 0.9565 - val_loss: 0.7599 - val_accuracy: 0.8380\n",
            "Epoch 33/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1223 - accuracy: 0.9577 - val_loss: 0.8067 - val_accuracy: 0.8296\n",
            "Epoch 34/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1124 - accuracy: 0.9605 - val_loss: 0.7600 - val_accuracy: 0.8392\n",
            "Epoch 35/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1092 - accuracy: 0.9615 - val_loss: 0.8057 - val_accuracy: 0.8347\n",
            "Epoch 36/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1118 - accuracy: 0.9614 - val_loss: 0.8527 - val_accuracy: 0.8311\n",
            "Epoch 37/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1066 - accuracy: 0.9620 - val_loss: 0.8275 - val_accuracy: 0.8366\n",
            "Epoch 38/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1000 - accuracy: 0.9647 - val_loss: 0.8123 - val_accuracy: 0.8407\n",
            "Epoch 39/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0969 - accuracy: 0.9665 - val_loss: 0.7878 - val_accuracy: 0.8391\n",
            "Epoch 40/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.1034 - accuracy: 0.9638 - val_loss: 0.8221 - val_accuracy: 0.8392\n",
            "Epoch 41/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0948 - accuracy: 0.9663 - val_loss: 0.8130 - val_accuracy: 0.8365\n",
            "Epoch 42/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0915 - accuracy: 0.9677 - val_loss: 0.8657 - val_accuracy: 0.8327\n",
            "Epoch 43/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0868 - accuracy: 0.9694 - val_loss: 0.8262 - val_accuracy: 0.8418\n",
            "Epoch 44/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0913 - accuracy: 0.9677 - val_loss: 0.8693 - val_accuracy: 0.8426\n",
            "Epoch 45/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0842 - accuracy: 0.9710 - val_loss: 0.8426 - val_accuracy: 0.8449\n",
            "Epoch 46/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0828 - accuracy: 0.9708 - val_loss: 0.8486 - val_accuracy: 0.8414\n",
            "Epoch 47/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0826 - accuracy: 0.9704 - val_loss: 0.8414 - val_accuracy: 0.8374\n",
            "Epoch 48/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0820 - accuracy: 0.9725 - val_loss: 0.8724 - val_accuracy: 0.8397\n",
            "Epoch 49/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0836 - accuracy: 0.9711 - val_loss: 0.8950 - val_accuracy: 0.8467\n",
            "Epoch 50/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0735 - accuracy: 0.9754 - val_loss: 0.9597 - val_accuracy: 0.8322\n",
            "Epoch 51/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0814 - accuracy: 0.9718 - val_loss: 0.8179 - val_accuracy: 0.8392\n",
            "Epoch 52/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0817 - accuracy: 0.9718 - val_loss: 0.9197 - val_accuracy: 0.8395\n",
            "Epoch 53/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0742 - accuracy: 0.9744 - val_loss: 0.8402 - val_accuracy: 0.8442\n",
            "Epoch 54/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0765 - accuracy: 0.9735 - val_loss: 0.8575 - val_accuracy: 0.8491\n",
            "Epoch 55/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0775 - accuracy: 0.9737 - val_loss: 0.8664 - val_accuracy: 0.8451\n",
            "Epoch 56/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0712 - accuracy: 0.9760 - val_loss: 0.8434 - val_accuracy: 0.8461\n",
            "Epoch 57/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0714 - accuracy: 0.9760 - val_loss: 0.9003 - val_accuracy: 0.8380\n",
            "Epoch 58/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0691 - accuracy: 0.9767 - val_loss: 0.8907 - val_accuracy: 0.8442\n",
            "Epoch 59/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0642 - accuracy: 0.9787 - val_loss: 0.9682 - val_accuracy: 0.8391\n",
            "Epoch 60/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0684 - accuracy: 0.9770 - val_loss: 0.8137 - val_accuracy: 0.8471\n",
            "Epoch 61/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0638 - accuracy: 0.9796 - val_loss: 0.8777 - val_accuracy: 0.8445\n",
            "Epoch 62/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0671 - accuracy: 0.9772 - val_loss: 0.9005 - val_accuracy: 0.8446\n",
            "Epoch 63/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0652 - accuracy: 0.9778 - val_loss: 0.9280 - val_accuracy: 0.8451\n",
            "Epoch 64/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0646 - accuracy: 0.9784 - val_loss: 0.9126 - val_accuracy: 0.8450\n",
            "Epoch 65/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0619 - accuracy: 0.9778 - val_loss: 0.8781 - val_accuracy: 0.8467\n",
            "Epoch 66/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0607 - accuracy: 0.9796 - val_loss: 0.9285 - val_accuracy: 0.8437\n",
            "Epoch 67/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0626 - accuracy: 0.9787 - val_loss: 0.9556 - val_accuracy: 0.8466\n",
            "Epoch 68/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0607 - accuracy: 0.9792 - val_loss: 0.9101 - val_accuracy: 0.8474\n",
            "Epoch 69/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0626 - accuracy: 0.9793 - val_loss: 0.8197 - val_accuracy: 0.8470\n",
            "Epoch 70/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0597 - accuracy: 0.9801 - val_loss: 0.9257 - val_accuracy: 0.8476\n",
            "Epoch 71/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0578 - accuracy: 0.9803 - val_loss: 0.9423 - val_accuracy: 0.8485\n",
            "Epoch 72/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0635 - accuracy: 0.9780 - val_loss: 0.8259 - val_accuracy: 0.8473\n",
            "Epoch 73/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0563 - accuracy: 0.9813 - val_loss: 0.9264 - val_accuracy: 0.8452\n",
            "Epoch 74/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0505 - accuracy: 0.9827 - val_loss: 1.0024 - val_accuracy: 0.8444\n",
            "Epoch 75/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0585 - accuracy: 0.9805 - val_loss: 0.9489 - val_accuracy: 0.8453\n",
            "Epoch 76/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0522 - accuracy: 0.9830 - val_loss: 0.9600 - val_accuracy: 0.8417\n",
            "Epoch 77/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0518 - accuracy: 0.9821 - val_loss: 0.9466 - val_accuracy: 0.8436\n",
            "Epoch 78/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0520 - accuracy: 0.9821 - val_loss: 0.9592 - val_accuracy: 0.8485\n",
            "Epoch 79/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0542 - accuracy: 0.9818 - val_loss: 0.9976 - val_accuracy: 0.8446\n",
            "Epoch 80/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0553 - accuracy: 0.9820 - val_loss: 0.9714 - val_accuracy: 0.8468\n",
            "Epoch 81/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0505 - accuracy: 0.9835 - val_loss: 0.9743 - val_accuracy: 0.8467\n",
            "Epoch 82/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0576 - accuracy: 0.9809 - val_loss: 0.9849 - val_accuracy: 0.8449\n",
            "Epoch 83/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0483 - accuracy: 0.9834 - val_loss: 0.9417 - val_accuracy: 0.8478\n",
            "Epoch 84/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0487 - accuracy: 0.9840 - val_loss: 0.9695 - val_accuracy: 0.8493\n",
            "Epoch 85/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0532 - accuracy: 0.9823 - val_loss: 0.9538 - val_accuracy: 0.8492\n",
            "Epoch 86/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0500 - accuracy: 0.9832 - val_loss: 0.9055 - val_accuracy: 0.8505\n",
            "Epoch 87/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0470 - accuracy: 0.9837 - val_loss: 0.9941 - val_accuracy: 0.8491\n",
            "Epoch 88/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0483 - accuracy: 0.9833 - val_loss: 0.9694 - val_accuracy: 0.8489\n",
            "Epoch 89/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0491 - accuracy: 0.9834 - val_loss: 0.9707 - val_accuracy: 0.8456\n",
            "Epoch 90/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0524 - accuracy: 0.9834 - val_loss: 0.9001 - val_accuracy: 0.8561\n",
            "Epoch 91/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0442 - accuracy: 0.9859 - val_loss: 0.9657 - val_accuracy: 0.8489\n",
            "Epoch 92/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0471 - accuracy: 0.9834 - val_loss: 0.9704 - val_accuracy: 0.8547\n",
            "Epoch 93/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0470 - accuracy: 0.9851 - val_loss: 0.9933 - val_accuracy: 0.8530\n",
            "Epoch 94/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0465 - accuracy: 0.9851 - val_loss: 0.9388 - val_accuracy: 0.8499\n",
            "Epoch 95/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0460 - accuracy: 0.9853 - val_loss: 0.9904 - val_accuracy: 0.8511\n",
            "Epoch 96/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0415 - accuracy: 0.9864 - val_loss: 0.9658 - val_accuracy: 0.8496\n",
            "Epoch 97/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0450 - accuracy: 0.9857 - val_loss: 0.9771 - val_accuracy: 0.8509\n",
            "Epoch 98/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0471 - accuracy: 0.9847 - val_loss: 1.0299 - val_accuracy: 0.8456\n",
            "Epoch 99/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0461 - accuracy: 0.9842 - val_loss: 0.9988 - val_accuracy: 0.8512\n",
            "Epoch 100/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0449 - accuracy: 0.9842 - val_loss: 0.9571 - val_accuracy: 0.8501\n",
            "Epoch 101/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0456 - accuracy: 0.9848 - val_loss: 0.9710 - val_accuracy: 0.8510\n",
            "Epoch 102/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0462 - accuracy: 0.9855 - val_loss: 0.9928 - val_accuracy: 0.8488\n",
            "Epoch 103/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0430 - accuracy: 0.9861 - val_loss: 1.0887 - val_accuracy: 0.8454\n",
            "Epoch 104/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0395 - accuracy: 0.9865 - val_loss: 0.9451 - val_accuracy: 0.8502\n",
            "Epoch 105/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0445 - accuracy: 0.9841 - val_loss: 0.9713 - val_accuracy: 0.8535\n",
            "Epoch 106/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0418 - accuracy: 0.9865 - val_loss: 1.0187 - val_accuracy: 0.8468\n",
            "Epoch 107/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0385 - accuracy: 0.9876 - val_loss: 1.0367 - val_accuracy: 0.8467\n",
            "Epoch 108/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0437 - accuracy: 0.9856 - val_loss: 0.9501 - val_accuracy: 0.8488\n",
            "Epoch 109/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0389 - accuracy: 0.9857 - val_loss: 1.0439 - val_accuracy: 0.8463\n",
            "Epoch 110/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0407 - accuracy: 0.9863 - val_loss: 1.0467 - val_accuracy: 0.8467\n",
            "Epoch 111/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0455 - accuracy: 0.9858 - val_loss: 0.9741 - val_accuracy: 0.8463\n",
            "Epoch 112/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0405 - accuracy: 0.9868 - val_loss: 0.9573 - val_accuracy: 0.8517\n",
            "Epoch 113/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0417 - accuracy: 0.9866 - val_loss: 0.9710 - val_accuracy: 0.8521\n",
            "Epoch 114/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0407 - accuracy: 0.9860 - val_loss: 1.0138 - val_accuracy: 0.8542\n",
            "Epoch 115/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0370 - accuracy: 0.9880 - val_loss: 1.0742 - val_accuracy: 0.8510\n",
            "Epoch 116/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0373 - accuracy: 0.9879 - val_loss: 0.9938 - val_accuracy: 0.8535\n",
            "Epoch 117/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0379 - accuracy: 0.9873 - val_loss: 1.0860 - val_accuracy: 0.8459\n",
            "Epoch 118/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0419 - accuracy: 0.9870 - val_loss: 0.9314 - val_accuracy: 0.8545\n",
            "Epoch 119/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0388 - accuracy: 0.9870 - val_loss: 1.0866 - val_accuracy: 0.8548\n",
            "Epoch 120/120\n",
            "391/391 [==============================] - 9s 22ms/step - loss: 0.0405 - accuracy: 0.9871 - val_loss: 1.0534 - val_accuracy: 0.8506\n",
            "Test loss:  1.053\n",
            "Test accuracy:  85.060\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}