CNN_models.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "group1_other_algorithms.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y3-k1kgExgdQ"
      },
      "source": [
        "# Import datasets & libraries\n",
        "from keras.datasets import cifar100\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "v7c0P3Ucx3l9"
      },
      "source": [
        "# 1. Exploring the CIFAR-100 Dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ljCsgwglxuYw",
        "outputId": "118e46b9-1c28-405f-a459-140475df90be"
      },
      "source": [
        "# Download dataset of CIFAR-100\n",
        "(x_train, y_train), (x_test, y_test) = cifar100.load_data()\n",
        "\n",
        "# Check the shape of the array\n",
        "print(f\"x_train shape: {x_train.shape}\")\n",
        "print(f\"y_train shape: {y_train.shape}\")\n",
        "print(f\"x_test shape: {x_test.shape}\")\n",
        "print(f\"y_test shape: {y_test.shape}\")\n",
        "\n",
        "# Number of samples in dataset\n",
        "print(f\"Train: {x_train.shape[0]}\")\n",
        "print(f\"Test: {x_test.shape[0]}\")\n",
        "\n",
        "# Data format\n",
        "print(type(x_train))\n",
        "print(type(y_train))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-100-python.tar.gz\n",
            "169009152/169001437 [==============================] - 4s 0us/step\n",
            "169017344/169001437 [==============================] - 4s 0us/step\n",
            "x_train shape: (50000, 32, 32, 3)\n",
            "y_train shape: (50000, 1)\n",
            "x_test shape: (10000, 32, 32, 3)\n",
            "y_test shape: (10000, 1)\n",
            "Train: 50000\n",
            "Test: 10000\n",
            "<class 'numpy.ndarray'>\n",
            "<class 'numpy.ndarray'>\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FArhNnKex7qq",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 594
        },
        "outputId": "51effc15-3b0e-4645-9c7f-8987fa0d410d"
      },
      "source": [
        "# Show randomly 16 images in the CIFAR-100\n",
        "plt.figure(figsize=(10, 10))\n",
        "for i in range (16):\n",
        "  rand_num = np.random.randint(0, 50000)\n",
        "  cifar_img = plt.subplot(4,4, i+1)\n",
        "  plt.imshow(x_train[rand_num])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 16 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LUfGlaqvyBFG"
      },
      "source": [
        "# 2. Preprocessing for CIFAR-100 dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ithOUBUeyDPn"
      },
      "source": [
        "# Import libraries for preprocessing images\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "\n",
        "# Normalize images\n",
        "train_images = x_train.astype('float32')/255\n",
        "test_images = x_test.astype('float32')/255\n",
        "\n",
        "# Transform labels to one hot encoding\n",
        "train_labels = to_categorical(y_train)\n",
        "test_labels = to_categorical(y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yQvO64qyyPGf"
      },
      "source": [
        "# 3. Build CNN models for classification task"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A17TMrXeyRgG"
      },
      "source": [
        "# Import Libraries for CNN\n",
        "from keras.models import Sequential\n",
        "from keras.layers import MaxPooling2D, Conv2D, Flatten, Dense, Activation, Dropout"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o1qm9jIA2jTJ"
      },
      "source": [
        "# Plot function for visualisation training process\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "def training_plot(history):\n",
        "  acc = history.history['acc']\n",
        "  val_acc = history.history['val_acc']\n",
        "  loss = history.history['loss']\n",
        "  val_loss = history.history['val_loss']\n",
        "  epochs = range(1, len(acc) + 1)\n",
        "  plt.plot(epochs, acc, 'bo', label='Training acc')\n",
        "  plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
        "  plt.title('Training and validation accuracy')\n",
        "  plt.legend()\n",
        "  plt.figure()\n",
        "  plt.plot(epochs, loss, 'bo', label='Training loss')\n",
        "  plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
        "  plt.title('Training and validation loss')\n",
        "  plt.legend()\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x7nqYCMJyYdt"
      },
      "source": [
        "## 3.1 A simple CNN model with RMSprop optimizer and Adam optimizer"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lB_ESv9lyrb_",
        "outputId": "0f37a085-7036-4471-976c-425baac5a730"
      },
      "source": [
        "# Model\n",
        "# Build CNN model\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(32,32,3)))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(64, activation='relu'))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 30, 30, 32)        896       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 15, 15, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 13, 13, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 6, 6, 64)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 4, 4, 64)          36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 2, 2, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 64)                16448     \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 100)               6500      \n",
            "=================================================================\n",
            "Total params: 79,268\n",
            "Trainable params: 79,268\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m7kScBguyyJm"
      },
      "source": [
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xuPiZTcAy-Iv"
      },
      "source": [
        "# Training and testing model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=400,\n",
        "                    validation_split=0.2) # 20% of training set is used for validation\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g6O78vgU3IOs",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "outputId": "8d2bd60a-6f59-48ee-bf46-d44f9d2df895"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Hy41iIITl7Jn",
        "outputId": "51c23eb5-4baa-4393-e5c9-ad3badd05f71"
      },
      "source": [
        "# Test Accuracy on test set\n",
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 5ms/step - loss: 2.6309 - acc: 0.3593\n",
            "accuracy on test set: acc of 35.92999875545502\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MAZOg8OoEyWp",
        "outputId": "8fd4da6a-e075-4c4c-c4f0-b9a7dc9e8015"
      },
      "source": [
        "# Using Adam optimizer\n",
        "\n",
        "# Build CNN model\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(32,32,3)))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(64, activation='relu'))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.Adam(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 30, 30, 32)        896       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 15, 15, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 13, 13, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 6, 6, 64)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 4, 4, 64)          36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 2, 2, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 64)                16448     \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 100)               6500      \n",
            "=================================================================\n",
            "Total params: 79,268\n",
            "Trainable params: 79,268\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ooddUTcrE7S-"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=400,\n",
        "                    validation_split=0.2) # 20% of training set is used for validation\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "We6f9wd3HM-9",
        "outputId": "59628aec-0cf5-4121-8317-50b786e43133"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RJ2P5goIHQP7",
        "outputId": "8390b4e5-dd49-49cd-ecc5-5eaf3a02d3e5"
      },
      "source": [
        "# Test Accuracy on test set\n",
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 5ms/step - loss: 2.6010 - acc: 0.3563\n",
            "accuracy on test set: acc of 35.62999963760376\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rMu63UBBRdI0"
      },
      "source": [
        "## 3.2 Increased filters in CNN and increased conv layers in CNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ky4ipHssRq9V",
        "outputId": "f85254e5-15de-495f-c272-1afeeabcac8a"
      },
      "source": [
        "# Build CNN model with increase of filters\n",
        "model = Sequential()\n",
        "model.add(Conv2D(64, (3,3), activation='relu', input_shape=(32,32,3)))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "\n",
        "model.add(Conv2D(128, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "\n",
        "model.add(Conv2D(128, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 30, 30, 64)        1792      \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 15, 15, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 13, 13, 128)       73856     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 6, 6, 128)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 4, 4, 128)         147584    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 2, 2, 128)         0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               262656    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 100)               51300     \n",
            "=================================================================\n",
            "Total params: 537,188\n",
            "Trainable params: 537,188\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A99xU4ZJSmlD"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=400,\n",
        "                    validation_split=0.2) # 20% of training set is used for validation\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "Oaj44hD2S3ey",
        "outputId": "c1eabc27-eff3-4b27-df9d-1b67e6d866c9"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "68-i5Uk-Sz1i",
        "outputId": "4dc3e4b7-f930-4562-87d4-04d780cc6300"
      },
      "source": [
        "# Test Accuracy on test set\n",
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 2s 6ms/step - loss: 7.0105 - acc: 0.3373\n",
            "accuracy on test set: acc of 33.730000257492065\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KOHk1iPdvRkf",
        "outputId": "25475048-d137-4db2-ae91-9ec63663f202"
      },
      "source": [
        "# Build CNN model with increase of convolution layers\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(32,32,3)))\n",
        "model.add(Conv2D(32, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "\n",
        "\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 30, 30, 32)        896       \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 28, 28, 32)        9248      \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 14, 14, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 12, 12, 64)        18496     \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 10, 10, 64)        36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 5, 5, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 1600)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               819712    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 100)               51300     \n",
            "=================================================================\n",
            "Total params: 936,580\n",
            "Trainable params: 936,580\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "stsNA6RkJqID"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=400,\n",
        "                    validation_split=0.2) # 20% of training set is used for validation\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d5gPY3NCU23i"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=400,\n",
        "                    validation_split=0.2) # 20% of training set is used for validation\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "cfJ2K1JzTAGD",
        "outputId": "b9525cfe-4cae-4b80-d6fb-4d8192d25ca9"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TMDSnsqDTUkT",
        "outputId": "c7cea3c9-8896-4093-9e56-d75fc59520b1"
      },
      "source": [
        "# Test Accuracy on test set\n",
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 2s 5ms/step - loss: 12.1721 - acc: 0.3017\n",
            "accuracy on test set: acc of 30.169999599456787\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8TlXpjxbu_n1"
      },
      "source": [
        "## 3.3 Simple model with data augmentation and simple model added dropout layers with data augmentation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hAT_Au77vKGZ",
        "outputId": "19c5477d-cf2b-49e6-f6cf-6adc018e80a3"
      },
      "source": [
        "# Build CNN model\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(32,32,3)))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(64, activation='relu'))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 30, 30, 32)        896       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 15, 15, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 13, 13, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 6, 6, 64)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 4, 4, 64)          36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 2, 2, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 64)                16448     \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 100)               6500      \n",
            "=================================================================\n",
            "Total params: 79,268\n",
            "Trainable params: 79,268\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MDoDxCQzv3gE"
      },
      "source": [
        "# Data Augmentation\n",
        "# Adding data augmentation for creating more images\n",
        "# Divide train and validation set \n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# Configuration for creating new images\n",
        "train_datagen = ImageDataGenerator(\n",
        "    rotation_range=20,\n",
        "    horizontal_flip=True,\n",
        ")\n",
        "\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(train_images, train_labels, test_size=0.2, random_state=93)\n",
        "train_datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g-aFv7Elv6UE"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_datagen.flow(X_train, y_train, batch_size=64),\n",
        "          steps_per_epoch=100, \n",
        "          epochs=400,\n",
        "          validation_data=(X_validation, y_validation),\n",
        "          verbose=1)\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "wPa7j-MmwWhT",
        "outputId": "7ba94eea-67ed-4055-9be2-fc5688ad009d"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vqKm98wzwdHt",
        "outputId": "62a16113-3c0c-48d6-d6f6-b7af5c0b6499"
      },
      "source": [
        "# Test Accuracy on test set\n",
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 4ms/step - loss: 2.8913 - acc: 0.2942\n",
            "accuracy on test set: acc of 29.420000314712524\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v1k7Qd4jw1J_"
      },
      "source": [
        "# CNN increased layers and nodes + Dropout with data augmentation\n",
        "model = Sequential()\n",
        "# Stack 1\n",
        "model.add(Conv2D(64, (3,3), activation='relu', input_shape=(32, 32, 3)))\n",
        "model.add(Conv2D(64, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "# Stack 2\n",
        "model.add(Conv2D(128, (3,3), activation='relu'))\n",
        "model.add(Conv2D(128, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dlPZTuA3xZJR"
      },
      "source": [
        "# Data Augmentation\n",
        "# Adding data augmentation for creating more images\n",
        "# Divide train and validation set \n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# Configuration for creating new images\n",
        "train_datagen = ImageDataGenerator(\n",
        "    rotation_range=20,\n",
        "    horizontal_flip=True,\n",
        ")\n",
        "\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(train_images, train_labels, test_size=0.2, random_state=93)\n",
        "train_datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OllpLmCsxkYr"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_datagen.flow(X_train, y_train, batch_size=64),\n",
        "          steps_per_epoch=100, \n",
        "          epochs=400,\n",
        "          validation_data=(X_validation, y_validation),\n",
        "          verbose=1)\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "xEuw9nt6jbFB",
        "outputId": "5c251958-6910-4ccd-b52a-e8af55552fcb"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEICAYAAABPgw/pAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZgU1dX48e9hGNYBlYG4AILGHcUBBhVRg0t+ryDBDaOmFZAoghti4r4RE0zexCRIIhBccGHyotGEuOCGoKgkkgERQTGiMgFFgUG2jLIM5/fHraKX6W1mepnuPp/n6ae7qm5Xna6B07fvvXVLVBVjjDG5r1m2AzDGGJMaltCNMSZPWEI3xpg8YQndGGPyhCV0Y4zJE5bQjTEmT1hCN1GJyIsiMjzVZbNJRFaJyBlp2K+KyCHe66kicmcyZRtwnICIvNLQOOPsd4CIrEn1fk3mNc92ACZ1RGRbyGIbYDtQ6y1fqaoVye5LVQemo2y+U9XRqdiPiHQHPgOKVXWXt+8KIOm/oSk8ltDziKqW+K9FZBVwuarOiSwnIs39JGGMyR/W5FIA/J/UInKziHwJTBeRfUTkeRFZLyJfe6+7hLzndRG53Hs9QkTeEpH7vLKficjABpY9SETmi8hWEZkjIg+IyIwYcScT489F5G1vf6+ISMeQ7ZeKSJWIVIvI7XHOz/Ei8qWIFIWsO1dElnqvjxORf4jIJhFZKyJ/FJEWMfb1qIj8ImT5Ru89X4jIyIiyZ4nIuyKyRURWi8j4kM3zvedNIrJNRPr55zbk/SeKyL9EZLP3fGKy5yYeETnSe/8mEVkuIkNCtg0SkQ+8fX4uIj/11nf0/j6bRGSjiLwpIpZfMsxOeOHYD+gAdANG4f72073lA4FvgD/Gef/xwEdAR+DXwMMiIg0o+2dgIVAKjAcujXPMZGL8EXAZ8B2gBeAnmKOAKd7+D/CO14UoVPUd4L/AaRH7/bP3uhYY532efsDpwFVx4saL4Uwvnu8DhwKR7ff/BYYBewNnAWNE5Bxv2yne896qWqKq/4jYdwfgBWCS99l+B7wgIqURn6HOuUkQczHwHPCK975rgQoROdwr8jCu+a4dcDQw11v/E2AN0AnYF7gNsHlFMswSeuHYDdytqttV9RtVrVbVZ1S1RlW3AhOA78V5f5WqPqiqtcBjwP64/7hJlxWRA4G+wF2qukNV3wKejXXAJGOcrqr/VtVvgKeAMm/9UOB5VZ2vqtuBO71zEMv/ARcDiEg7YJC3DlVdpKr/VNVdqroK+FOUOKL5oRffMlX9L+4LLPTzva6q76vqblVd6h0vmf2C+wL4WFWf8OL6P2AF8IOQMrHOTTwnACXAr7y/0VzgebxzA+wEjhKR9qr6taouDlm/P9BNVXeq6ptqE0VlnCX0wrFeVb/1F0SkjYj8yWuS2IL7ib93aLNDhC/9F6pa470sqWfZA4CNIesAVscKOMkYvwx5XRMS0wGh+/YSanWsY+Fq4+eJSEvgPGCxqlZ5cRzmNSd86cVxL662nkhYDEBVxOc7XkTmeU1Km4HRSe7X33dVxLoqoHPIcqxzkzBmVQ398gvd7/m4L7sqEXlDRPp5638DrAReEZFPReSW5D6GSSVL6IUjsrb0E+Bw4HhVbU/wJ36sZpRUWAt0EJE2Ieu6xinfmBjXhu7bO2ZprMKq+gEucQ0kvLkFXNPNCuBQL47bGhIDrtko1J9xv1C6qupewNSQ/Saq3X6Ba4oKdSDweRJxJdpv14j27z37VdV/qerZuOaYWbiaP6q6VVV/oqoHA0OAG0Tk9EbGYurJEnrhaodrk97ktcfene4DejXeSmC8iLTwanc/iPOWxsT4NDBYRE7yOjDvIfG/9z8DY3FfHH+JiGMLsE1EjgDGJBnDU8AIETnK+0KJjL8d7hfLtyJyHO6LxLce10R0cIx9zwYOE5EfiUhzEbkQOArXPNIY7+Bq8zeJSLGIDMD9jWZ6f7OAiOylqjtx52Q3gIgMFpFDvL6Szbh+h3hNXCYNLKEXrolAa2AD8E/gpQwdN4DrWKwGfgE8iRsvH02DY1TV5cDVuCS9Fvga12kXj9+GPVdVN4Ss/yku2W4FHvRiTiaGF73PMBfXHDE3oshVwD0ishW4C6+26723Btdn8LY3cuSEiH1XA4Nxv2KqgZuAwRFx15uq7sAl8IG48z4ZGKaqK7wilwKrvKan0bi/J7hO3znANuAfwGRVndeYWEz9ifVbmGwSkSeBFaqa9l8IxuQ7q6GbjBKRviLyXRFp5g3rOxvXFmuMaSS7UtRk2n7AX3EdlGuAMar6bnZDMiY/WJOLMcbkCWtyMcaYPJG1JpeOHTtq9+7ds3V4Y4zJSYsWLdqgqp2ibctaQu/evTuVlZXZOrwxxuQkEYm8QngPa3Ixxpg8YQndGGPyhCV0Y4zJEzYO3ZgCsnPnTtasWcO3336buLDJqlatWtGlSxeKi4uTfo8ldGMKyJo1a2jXrh3du3cn9v1JTLapKtXV1axZs4aDDjoo6fflVJNLRQV07w7NmrnnCrtdrjH18u2331JaWmrJvIkTEUpLS+v9SypnaugVFTBqFNR4t0aoqnLLAIFA7PcZY8JZMs8NDfk75UwN/fbbg8ncV1Pj1htjjMmhhP6f/9RvvTGm6amurqasrIyysjL2228/OnfuvGd5x44dcd9bWVnJddddl/AYJ554Ykpiff311xk8eHBK9pUpOZPQD4y8eVeC9caYxkt1v1VpaSlLlixhyZIljB49mnHjxu1ZbtGiBbt27Yr53vLyciZNmpTwGAsWLGhckDksZxL6hAnQpk34ujZt3HpjTOr5/VZVVaAa7LdK9WCEESNGMHr0aI4//nhuuukmFi5cSL9+/ejVqxcnnngiH330ERBeYx4/fjwjR45kwIABHHzwwWGJvqSkZE/5AQMGMHToUI444ggCgQD+7LKzZ8/miCOOoE+fPlx33XUJa+IbN27knHPOoWfPnpxwwgksXboUgDfeeGPPL4xevXqxdetW1q5dyymnnEJZWRlHH300b775ZmpPWBxJd4p6d1qvBD5X1cER20bg7vrt36D2j6r6UKqChGDH509+Al99BfvuC7/9rXWIGpMu8fqtUv3/bs2aNSxYsICioiK2bNnCm2++SfPmzZkzZw633XYbzzzzTJ33rFixgnnz5rF161YOP/xwxowZU2fM9rvvvsvy5cs54IAD6N+/P2+//Tbl5eVceeWVzJ8/n4MOOoiLL744YXx33303vXr1YtasWcydO5dhw4axZMkS7rvvPh544AH69+/Ptm3baNWqFdOmTeN//ud/uP3226mtraUm8iSmUX1GuYwFPgTax9j+pKpe0/iQYgsE4KijoHdv+NOf4Oyz03k0YwpbJvutLrjgAoqKigDYvHkzw4cP5+OPP0ZE2LlzZ9T3nHXWWbRs2ZKWLVvyne98h6+++oouXbqElTnuuOP2rCsrK2PVqlWUlJRw8MEH7xnfffHFFzNt2rS48b311lt7vlROO+00qqur2bJlC/379+eGG24gEAhw3nnn0aVLF/r27cvIkSPZuXMn55xzDmVlZY06N/WRVJOLiHQBzgJSWutuiPbe18mWLdmNw5h8l8l+q7Zt2+55feedd3LqqaeybNkynnvuuZhjsVu2bLnndVFRUdT292TKNMYtt9zCQw89xDfffEP//v1ZsWIFp5xyCvPnz6dz586MGDGCxx9/PKXHjCfZNvSJuLuK745T5nwRWSoiT4tI18aHFt0rr7jnYcPs4iJj0ilb/VabN2+mc+fOADz66KMp3//hhx/Op59+yqpVqwB48sknE77n5JNPpsJLNq+//jodO3akffv2fPLJJxxzzDHcfPPN9O3blxUrVlBVVcW+++7LFVdcweWXX87ixYtT/hliSZjQRWQwsE5VF8Up9hzQXVV7Aq8Cj8XY1ygRqRSRyvXr19c72IoK14buS1cnjTHGNXFOmwbduoGIe542Lf39VjfddBO33norvXr1SnmNGqB169ZMnjyZM888kz59+tCuXTv22muvuO8ZP348ixYtomfPntxyyy089phLcRMnTuToo4+mZ8+eFBcXM3DgQF5//XWOPfZYevXqxZNPPsnYsWNT/hliSXhPURH5JXApsAtohWtD/6uqXhKjfBGwUVXjnqHy8nKt7w0uund3STxSt27gfdkaY+L48MMPOfLII7MdRtZt27aNkpISVJWrr76aQw89lHHjxmU7rDqi/b1EZJGqlkcrn7CGrqq3qmoXVe0OXATMjUzmIrJ/yOIQXOdpytnFRcaYVHjwwQcpKyujR48ebN68mSuvvDLbIaVEg+dyEZF7gEpVfRa4TkSG4GrxG4ERqQkv3IEHRq+h28VFxpj6GDduXJOskTdWvS4sUtXX/THoqnqXl8z9WnwPVT1WVU9V1RXpCNYuLjLGmNhy5kpRCHbS+CORMtVJY4wxuSBnps/1BQIwcyZ88QUsijfuxhhjCkxO1dB9GzfC0qV2owtjjAmVcwm9ogIWLoRdu9I7YZAxJvVOPfVUXn755bB1EydOZMyYMTHfM2DAAPwhzoMGDWLTpk11yowfP5777rsv7rFnzZrFBx98sGf5rrvuYs6cOfUJP6qmNM1uziX02293yTyU3ejCmNxw8cUXM3PmzLB1M2fOTGqCLHCzJO69994NOnZkQr/nnns444wzGrSvpirnErqNRTcmdw0dOpQXXnhhz80sVq1axRdffMHJJ5/MmDFjKC8vp0ePHtx9991R39+9e3c2bNgAwIQJEzjssMM46aST9kyxC26Med++fTn22GM5//zzqampYcGCBTz77LPceOONlJWV8cknnzBixAiefvppAF577TV69erFMcccw8iRI9m+ffue491999307t2bY445hhUr4g/gy/Y0uznXKWpj0Y1JjeuvhyVLUrvPsjKYODH29g4dOnDcccfx4osvcvbZZzNz5kx++MMfIiJMmDCBDh06UFtby+mnn87SpUvp2bNn1P0sWrSImTNnsmTJEnbt2kXv3r3p06cPAOeddx5XXHEFAHfccQcPP/ww1157LUOGDGHw4MEMHTo0bF/ffvstI0aM4LXXXuOwww5j2LBhTJkyheuvvx6Ajh07snjxYiZPnsx9993HQw/FnqMw29Ps5lwNfcIEaN06fJ2NRTcmd4Q2u4Q2tzz11FP07t2bXr16sXz58rDmkUhvvvkm5557Lm3atKF9+/YMGTJkz7Zly5Zx8sknc8wxx1BRUcHy5cvjxvPRRx9x0EEHcdhhhwEwfPhw5s+fv2f7eeedB0CfPn32TOgVy1tvvcWll14KRJ9md9KkSWzatInmzZvTt29fpk+fzvjx43n//fdp165d3H0nI+dq6P6Y80svdZ2iUDfBG2MSi1eTTqezzz6bcePGsXjxYmpqaujTpw+fffYZ9913H//617/YZ599GDFiRMxpcxMZMWIEs2bN4thjj+XRRx/l9ddfb1S8/hS8jZl+95ZbbuGss85i9uzZ9O/fn5dffnnPNLsvvPACI0aM4IYbbmDYsGGNijXnaug+keDr6mob6WJMrigpKeHUU09l5MiRe2rnW7ZsoW3btuy111589dVXvPjii3H3ccoppzBr1iy++eYbtm7dynPPPbdn29atW9l///3ZuXPnnilvAdq1a8fWrVvr7Ovwww9n1apVrFy5EoAnnniC733vew36bNmeZjfnaujgRrTsjpiZPV23xjLGpN7FF1/Mueeeu6fpxZ9u9ogjjqBr1670798/7vt79+7NhRdeyLHHHst3vvMd+vbtu2fbz3/+c44//ng6derE8ccfvyeJX3TRRVxxxRVMmjRpT2coQKtWrZg+fToXXHABu3btom/fvowePbpBn8u/12nPnj1p06ZN2DS78+bNo1mzZvTo0YOBAwcyc+ZMfvOb31BcXExJSUlKboSRcPrcdGnI9Lm+Zs2CzS2hROomemNMkE2fm1tSPn1uU5TJW2MZY0yuyMmEPmECRNzc20a6GGMKXk4m9EAABgwILhcVwfDh1n5uTDKy1cxq6qchf6ecTOgVFRAyTJTaWnjsMRvlYkwirVq1orq62pJ6E6eqVFdX06pVq3q9Lyc7Re3eosY0zM6dO1mzZk2Dx3ibzGnVqhVdunShOKJ9OV6naE4OW7T5XIxpmOLiYg466KBsh2HSJCebXGyUizHG1JWTCT3avUVFYNCg7MRjjDFNQU4m9EDAjWoJpWodo8aYwpaTCR1g9uy66+xGF8aYQpazCd06Ro0xJlzSCV1EikTkXRF5Psq2liLypIisFJF3RKR7KoOMxjpGjTEmXH1q6GOBD2Ns+zHwtaoeAvwe+N/GBpbIhAngTVO8h3WMGmMKWVIJXUS6AGcBse69dDbwmPf6aeB0kdAZy1MvEICLLgpfZx2jxphClmwNfSJwExBrctrOwGoAVd0FbAZKIwuJyCgRqRSRyvXr1zcg3HDz5tVdZx2jxphClTChi8hgYJ2qLmrswVR1mqqWq2p5p06dGrs7Vq+Ovt46Ro0xhSiZGnp/YIiIrAJmAqeJyIyIMp8DXQFEpDmwF1Cdwjijso5RY4wJSpjQVfVWVe2iqt2Bi4C5qnpJRLFnAf9Sn6FembTP+jVhAjSPmI3GOkaNMYWqwePQReQeERniLT4MlIrISuAG4JZUBJdIIACRtx60jlFjTKHKyelzQ+27L6xbV3d9aSls2NDo3RtjTJOSd/cUDRUtmQNUV1st3RhTWHI+ocfrALXhi8aYQpLzCf3ee2Nvs+GLxphCkvMJPRCoO9LFZ8MXjTGFJOcTOsCRR0Zfb8MXjTGFJC8S+pdfRl8/bZp1jBpjCkdeJPRY08LU1sKoUZbUjTGFIS8SerdusbfV1MDYsZmLxRhjsiUvEvqECfG325h0Y0whyIuEHggkLmNj0o0x+S4vEjpA167xt9uYdGNMvsubhP7LX0KLFrG325h0Y0y+y5uEHgjAI49AmzZ1t7Vpk7id3Rhjcl3eJHRwSf3JJ93rZt4nKyqC4cOTa2c3xphcllcJHWD5cve827v7aW0tTJ0KV12VvZiMMSYT8i6hP/BA3XWqLqnb0EVjTD7Lu4S+Zk309ap2gZExJr/lXUKPN5rFLjAyxuSzvEvoiUazWC3dGJOv8i6hBwJw3nmxt1st3RiTr/IuoQM89RSIxN5utXRjTD7Ky4ReVAT9+sXeXl2duViMMSZT8jKhA1x/ffzt1uxijMk3CRO6iLQSkYUi8p6ILBeRn0UpM0JE1ovIEu9xeXrCTd7gwfGbXUaOtKRujMkvydTQtwOnqeqxQBlwpoicEKXck6pa5j0eSmmUDdC6NQwZEnv7jh02pa4xJr80T1RAVRXY5i0Wew9NZ1CpMmtW/Fp6VVXmYjHGmHRLqg1dRIpEZAmwDnhVVd+JUux8EVkqIk+LSNTZyUVklIhUikjl+lg3Ak2x9u1jbxOxZhdjTP5IKqGraq2qlgFdgONE5OiIIs8B3VW1J/Aq8FiM/UxT1XJVLe/UqVNj4k7aj34Ue5uqNbsYY/JHvUa5qOomYB5wZsT6alXd7i0+BPRJTXiNd+WV8bdbs4sxJl8kM8qlk4js7b1uDXwfWBFRZv+QxSHAh6kMsjGOOQaOjvw9EcGaXYwx+SCZGvr+wDwRWQr8C9eG/ryI3CMi/jiS67whje8B1wEj0hNu/RUVwaRJ8ctcconNl26MyX3iBrFkXnl5uVZWVmbseAceCKtXx94uAk88YXc2MsY0bSKySFXLo23L2ytFI/3yl/G3WwepMSbXFUxCDwSgtDR+GesgNcbksoJJ6AD33w/FxfHLWAepMSZXFVRCDwRg+vT4ZazZxRiTqwoqoYNL6m3bxt5eVWW1dGNMbiq4hA4wdGj87TYTozEmFxVkQv/jH+Nv37HD7mpkjMk9BZnQS0rgn/+MX8buPWqMyTUFmdABjj8eunWLX8Y6SI0xuaRgEzrAhAk2X7oxJn8UdEIPBGDQoPhlbI4XY0yuKOiEDvDII/G3T5kC7dpZe7oxpukr+IT+ne+4ibvi2bbNhjIaY5q+gk/oAPfem7iM3VTaGNPUWULHtaWffnricv/5T/pjMcaYhrKE7pk1Czp2jF+mQ4fw5a1bobY2fTEZY0x9WEL3lJTA++/HL1NdDa1awQMPwK5d0L69XVFqjGk6LKGH2G8/uPxyd9u6WLZvh2uuceUg8SgZY4zJFEvoER58EB57LHE5v0yrVumNxxhjkmUJPYpAIHF7uq916/TGYowxybKEHsOvf51cuV270huHMcYkyxJ6DJddlvh2dQDr1tkFR8aYpiFhQheRViKyUETeE5HlIvKzKGVaisiTIrJSRN4Rke7pCDbTpk9PLqlfconN+WKMyb5kaujbgdNU9VigDDhTRE6IKPNj4GtVPQT4PfC/qQ0zO/x7kCaaZhfcnC9nnJH+mIwxJpaECV2dbd5isffQiGJnA/7YkKeB00XiTUybOwIBWLUKSksTl33tNaupG2OyJ6k2dBEpEpElwDrgVVV9J6JIZ2A1gKruAjYDdVKgiIwSkUoRqVy/fn3jIs+w++9PrtyUKdC9u7WrG2MyL6mErqq1qloGdAGOE5GjG3IwVZ2mquWqWt6pU6eG7CJrAgEYMya5slVVrlPVkroxJpPqNcpFVTcB84AzIzZ9DnQFEJHmwF5AdSoCbEomT3Y18GTs3GnTAhhjMiuZUS6dRGRv73Vr4PvAiohizwLDvddDgbmqGtnOnhdGj4Z581yzSiLV1db8YozJnOZJlNkfeExEinBfAE+p6vMicg9QqarPAg8DT4jISmAjcFHaIm4CBgyAzz5zl/1v3x6/bFUVjBrlXgcCaQ/NGFPAEiZ0VV0K9Iqy/q6Q198CF6Q2tKZv+HCYNi1xuZoaGDbMvT74YOjZE9q2TW9sxpjCY1eKNsIDD8DJJydXdvdudwHSiSfCkCHpjcsYU5gsoTdC8+Ywfz7MmAH1GXU/d276YjLGFC5L6CkQCMDvf1+/96xenZ5YjDGFyxJ6iowd60bAJOu3v01fLMaYwmQJPYXqM5/L/fcHpwn43e/gjjvSF5cxpjBYQk+xV191berJzP0yZYpre//JT2DCBPj4Y7joIujcOf1xGmPyj2Tr+p/y8nKtrKzMyrEzpXt3Nw69IfLzsixjTGOJyCJVLY+2zWroaTRhQnLzqUezeXNqYzHG5D9L6Gnkz6feEFOnpjYWY0z+s4SeZoFAcjfIiHT33TYHjDGmfiyhZ8CECdCmTf3es3073HpreuIxxuQnS+gZEAi4OV/8mnqyV5WuXg0tWlhN3RiTHEvoGeLfyk7VzeuS7M0ydu50E3tZUjfGJGIJPUsmT4aSkuTK+hN7tW9vid0YE5sl9CyaOrV+k3pt3Rq8td1nnzV8jLsxJj9ZQs+iQKB+87+Aa4K55BI3r3rv3q4Jp6YmvMxf/wp5fs2WMSaKZO5YZNJo8mT3nOy9SkNt3Oim8G3VCtasgX32cevPP98929WmxhQWq6E3AZMnJz//S6Tdu10NvUMHN9XAQw/VLaMKy5Y1OkxjTBNnCb2JCARgwwaXfJMdAROpqgquvbbu+hkz4Jhj4IUXYtfav/kGdu1q2HGNMU2DJfQmaPLkhtXWAb79Nvh6zhz46CN47jm3PHgwnHRS9Pe1aQM/+EHDjmmMaRosoTdR999f/6tLI33/+3DEEfCXvwTXLVjgRtf85z/B2nptrXt+6aXGHc8Yk12W0Juohl5dmowxY9x+zznHDX+85prgtt27U3ccY0xmWUJvwiKvLp0xo/G1dl9RETz7LPTsGT6z4yefuCT/9dduWRXuu8/dfMMY07QlTOgi0lVE5onIByKyXETGRikzQEQ2i8gS73FXesItbH6tvaHt6y1aBF/7zSzbtoWX+d733Bj3q692y199BTfeCEOGuPumhtbmjTFNSzI19F3AT1T1KOAE4GoROSpKuTdVtcx73JPSKM0e/miY00+v/3t37Ii//cADYe1a93rOHPf82Wfuef16V6N/9tn6H9cYkxkJE7qqrlXVxd7rrcCHgN31MsvmzHFt4alsW1+4EB5+GO691yXwDRvg3/9223budMMiV6+uW6v3qboavTEmO+rVhi4i3YFewDtRNvcTkfdE5EUR6RHj/aNEpFJEKtevX1/vYE24yZNd23qyk3wlMmcOjBwJJ5zgljt1ghEj3OstW4KjYlasCL5n61Z4/nn3esoU2G8/N1TSGJN5SSd0ESkBngGuV9UtEZsXA91U9VjgD8CsaPtQ1WmqWq6q5Z06dWpozCbC1Knh7eMNdfvt7rm83F15GsuHHwZfDxzoxq+vXAl//7tb59fq6+Oaaxo2/YExJiiphC4ixbhkXqGqf43crqpbVHWb93o2UCwiHVMaqYkpEIBHHml4Z6nPn72xXTs3N8yqVXXLtG4Njz4Kw4fDUUfB22+79YsWBYc8JnuDa1U3cqemBh54AK66qnHxG1PokhnlIsDDwIeq+rsYZfbzyiEix3n7rU5loCY+v7N0xoyG3cPUJ+Iebdq4uWFKSuDMM922iy6CiRNh7lx4/PHwmvrixcGEvmZNcsd64w249NLo0xUYY+ovmRp6f+BS4LSQYYmDRGS0iPiTvw4FlonIe8Ak4CJVm+svG/yx6zNmQHFx4/e3bRu89prrgH3sMdfGHuqQQ6BPHzdd78aNbt2CBa5T9c034Q9/CJZdsADmzw8u+yNq3noruM4fThlLbS1MmlT3V8CyZe7qV2MKmqpm5dGnTx816TVjhmppqapr3Gjco7Q0uN/Q9XfeqXrzzaoiqi1aRH/vSy+p7r9/cLlrV9WlS1Xvvdctl5QEt1VV1f0ctbWqM2eq7tqlOmeOK3fOOeFlwMVQH9u2qR53nOo//lH/c2tMtgCVGiOv2pWieSx0BscZM6Bt24bvq7o62MY9YwZ06eJGvowfD3feCYcdVnecu3+8M890tXG/43b1aneF6m23ueXQYZD+uPdQjz/umnumTIEPPnDrZs2C7dvd6y1eF319fxP++99uqGborwZjcpkl9AIRCLjE2ZALknxTp7rb3wUCLim3awfNmrnEPXMm7LVXsOy6deHNIjNnugT86qtuaGMs773nnkOT85tK6y4AABdLSURBVCefuOcvvoD33w+uX7TIfbH8LqRnx0/yyfj88/BnY3KdJfQCM2dOwztOVd3l/xUVrsO0WTP3XFEBZWWuFl9V5drOO3Vy88X8/OeuBn/hhW4fZ5zhRskcckj4vps3d+Pf773X1ejPOccd76WXgjfG3rwZli4Nzmfzt7+5ZPyznwX3U585Z/xE/sUX9T8XxjRJsdpi0v2wNvTsmzFDtU2bxrevt2nj9uXvs1s3157drVtwfTRvvaX69NOqRx6peuONqi+/XHe/ocunnKLaurXqj37klvv2rRvLX/6S3Gd/6qnge048sWHnb+dO1csvV/3gg4a935iGIE4buiX0AjdjhmpRUeOTup+8I5NwaLJPZMuW4Pv691c96aTox5o9W7Vt2+DyoYeqTp3qvkRGjlSdMkV17VrX6dquner69ao7doQfK3R/3buHb1u4UHXx4sTxLlrk3l9e7pL6p58m9zmNaQxL6CauVNXU4yX7ZPnv+fBDt7x7t+rYsaqnnebWN2+u+u23qscc45b32y/43u7dg+//7nfDYzjiCNX58+seB9zonN27625LZPZsV86PBdyX48qVyX/exth/f9Xvfz8zxzJNR7yEbm3oJuxmGiLuecaM1O2/PuPDe/Vyz4cd5p5F3MVM06fDZZfBr38NLVvCd7/rtofOIHHEEe65WbNgR6pv7Vo45RT46U/d/VND7dgRHENfHXI53JIl8Oc/x47Vv7K2Wcj/otpaeOaZxJ+zsVTdZ3r11dhl3nor8bj+eL76yt05S+s5eqiQ1dYGO/bB/VsIBDIYQKxMn+6H1dCbvlSNYW/bNvljfv216kcfJS43b57b90knBdddcYVbd+edrjZ+993BGNavV73ySvf6+uuDtXj/PQ8+6Grpv/td3fiXLYsew803u+2dOoWXHzIk+Gtg3TrVNWuS++yPPJJ8s8369fF/Sfi/HiZNSm5/0ZxxhtvH4sWuv6DQVFeH/3JLxu23u3O2fLlb9v9GtbWpiwtrcjENkar29ch29lR55RXVjz8OLl93nTvOww8H1115perFF7vXtbWqJ5wQjGfxYvcftn9/1X32cZ2joW3//ut+/Vyb/HPPqT7wgGubX7dO9cIL43/em292TUVHHhmeGJYscR28mzYF161aFXzfl1+qfvKJ6sCB4Z8v1D//GSz/n/+ovvpq+HY/sVx5ZcPP70EHBY/x//5fw/eTaxYsUP3Nb9zn/sUv6vfeXr3c+956yy375+/rr1MXnyV002CpvNrUf5x+enpiXbfOtbfX1MQuU1HhYthnH9cWr6r64ovB2O66y9WuA4HwmEeODF++6qroo2xiPfwa2/btqoMHu3UXXeQ6VlVVp0+P/r4f/zj+5wDVvfZyz6FfEKee6tYNHOg+5/XXuxFA0b5QlyxRnTCh7vrQq3tj/RIIVVur+uc/u8/YlNXWulFOsWrNkX+DIUOS37ffj/PSS+H7+uSTYJn77lN9442Gx28J3aRMKmvtRUWqY8ZkNv4dO1T/8AfVzz8PrqutVW3ZUvXgg4M16U8/dTX2zz5zHY+xPsOxx0Zf//TT4csTJrjjhI7O8R8XX+xqwB071t3WqpVrXon0s5/VLTt7ttu2aJHrPAbVAw5QnTUrvNx//xu+r7Fj3fotW8LX77133YS+e7f7lRKtGWnqVFfuj3+s/9/liy+iT/uQrCFDVA8/PLmyDz3k4pwyxX2eF18MNim99FL0v6ef/D//PPzfTqT27V35mTPdsv/+ykq3vGuXarNmqnfc0bDP6fZpCd2k0IwZ7h9lKmvsob8CSktT2zSTjHXrwmu4oSZPDsa2337B1337ul8D555bN1FXVQXntunUyQ1tnD8//nm44orobfi33uq++N5/39Wmv/wy+NM+9NGmjWqPHqqHHebivPNOt75///ByL78c/vnOPdetX7AgmNhqa1WLi8Pft2GD6rvvutf9+gXf7//S+cEP3LZ77419nh9/3H3BRPJrtu++685D6K+s3bvdF9hnn7nlL790+9m1K1gm2V8Rqqr33OPK3nab6v33u9cPPRTsl4n2qK52X4QQfdTWf/+runFjsLz/ZeEvz5ypOmKEG73V0C+94Ge1hG5SbMaM6LXNVD1atMh8Uo9lzRpX+5s719Vir7km+J/Ut22bW9ejh+pll7n/zPPmqZ53nmuHBTde3v98HTqonnVW+Jfa5s3ufV984da1bBn/18HEifG/HDZsCNYYW7YMfsGMGeMS9saNLsmEvu9Xv3KfJ7RNP3K//uuTT3Zt+cXFLkmH/nsYOjT6ufS3f/ONW/7FL4I1ZnAXjoGrwe7c6ZL4ihVuXc+ebiI1v+wrr7jP6P/CAPeZEvn5z13Zn/5UtXNn9/q664Lt5tEeH30U/mW7e7fq3/4W/HIM7ZsB1V/+0n0J+Mv+l90tt7jnp59u6L9GS+gmA7p1S09i9ztS63MFarpVV6v+6U9122DXrYs+GuSTT1yiKi8PJuG//c11lMWqrd1zj0teoZ2fF17o2sLLytzzjh2uY/fxx92+Vq4Mlp061e1n1izVceOCTSR+X0CPHtHPd8+erj1/8OC6NXSo29zm1+7BteX7XyAQrGV/843q+eervv12cNvjj7vzEuvvvvfe7toBCG9e6tnTNUP5561fv/D3Rf5dNmwIjlL6979VV69WHT/elT3kkOD7TjxR9YILYsfz9tuuz8Nf9mcILSpy+44sf8017m/jL/vnpawsuL+GsoRu0i6dFycVF9edmrc+V6A2BaFJJnTESzLD2aZPD3aeJuKfH7/NNtKOHS65JnPe/ZEy0db/9a91148YEeyMBdXXX3dNKM8/75b9RAyqo0erTpvWsH8P99zjfg0MGxZ9e48ebnrld991UzWD64j0v3QuvzxYtmtX1Wuvda9btqy7r4UL3fPf/x5MxpFfbps31/23megzhHaS1pcldJMRobXo0tLY86On6lFUlFtJPRP+8Ad3bvx27Wh27nRfEBs2BMeaR3u8/LJrOjjvPNcR6v8Ke/tt92sksvzEibE7iUMfobX4Aw4IvvaHSYZe5etfOxD6eO01d5xETX5HHRW+7Cdsv8MY3K+ehQtV+/RxHavt2gW3/eAHwaanadPcr6zQq4L//nf3/MYb4cc5+ODE5yCyY7o+LKGbrAhN8KnsRI2sDVlSb7jf/z72ua2uDi977bWq++4bbFbad9/w2ugbb6hefbV77bcpH3lk3f1edlnw9Z/+FHztDxUNbcZZsqTu+7dudW30oYkXwodZXnJJ9C+C0C8QcB3HofzOZn9Yod838sMfuucbbwy+t6oqfF/nnuuGp/rDSCG8UhNaw28MS+gm69LZJFOfuWJMuNpaN9xx82b32LAhdtLZti18yN64ce7iKb9zd9Mm98tg2TLXrPTpp+554kTXTNKhg+v8/dWvXPlrr3X78b/sn3vOPT/1lOr3vufaznfvdsM5hw1zyfKJJ9x77rgjGOeOHcErbDdtUn3ySTcW/rrrwi8W27pV9dJLw//t+Pvz+TN5vvtucF1oeb823q2biy30S+TFF+uW978IwHWsWkI3ecOvsacjqWdjqGO+WrnSDZFM1ubNrtkiWWvXurH3Gza45Zkz3VDLnTujX1G5a1fdS/A3bHAdt+PGxT9W6NBBVTcVQui/m3feCS+/ZYvrsA0Vmvx373YdzatXB8v72/3OV3/5uefcl+AddwR/CfTrpzpoUPyYE7GEbpqcdAx7bN7cDclrKqNhTNMwfrxrA1d1Y+1D/80kM8zxgAPcqJtY/vEP9yvFv0L2ttvce6Kp79ww0cRL6OK2Z155eblWVlZm5dim6bnqKneLu3T9cywqcjPhdesGEyZkeAY802R8842749WFF7rHuecmfs+2bW5GTf9OWdkmIotUtTzaNps+1zQJkyfDE0807NZ4yfCnka2qgksucdPyikDHjsFb3Jn817q1u3H6jBnJJXOAkpKmk8wTsYRumoxAAFatcrX0GTOgtDT9x6yudgneEnvhKC1197DNRwkTuoh0FZF5IvKBiCwXkbFRyoiITBKRlSKyVER6pydcUygCgWBNKhO1o+pqGDXKNf1E3gDbmFyRzPfULuAnqrpYRNoBi0TkVVX9IKTMQOBQ73E8MMV7NqZR/Lbu2293dz5q0wb++9/0HKumBqZMCS5XVbkkHxqHMU1Zwhq6qq5V1cXe663Ah0DniGJnA/5gn38Ce4vI/imP1hQkvylm927XQTVjRvra2iPV1LgmmebNg23uHTtaDd40TfVqQxeR7kAv4J2ITZ2B1SHLa6ib9BGRUSJSKSKV69evr1+kxniy0dbud6pWV7uHqqvBjxxpSd00HUkndBEpAZ4BrlfVLQ05mKpOU9VyVS3vFHp3X2MayG9r90cWZ6rN3bdjB4yt06tkTHYkldBFpBiXzCtU9a9RinwOdA1Z7uKtMyajAgGYNs01yYi453TX4qurg80vFRXWqWqyJ+GFRSIiwGPARlW9PkaZs4BrgEG4ztBJqnpcvP3ahUUmkyoq4LLLYOfO7By/WTPXB2AXNpnGauyFRf2BS4HTRGSJ9xgkIqNFZLRXZjbwKbASeBC4KhWBG5MqgQBMn56Z9vZodu92z/7IGRseadLBLv03BauiwiXXmppsR+La/adNs5q7Scwu/TcmitD2dnBt7tlSUwPDhlmN3TSOJXRT0EKHQPpzyYi4pplmGf7fsXt3cDikP9+MJXdTH5bQjfGEXsC0YQM8/njdIZAicNRRmYspXpu7jagxkawN3Zg4KiqC0w4ceGBwhEpFhWsi8Ts7M6242H257NgRXGft8IUhXhu6JXRjGqgpdar6iorcl0zol4/JL9YpakwaRF7EVFrqHqEXNGVqagJfbW2wHf7SS11TjSkcltCNaYTIdvcNG9zrVavctsipCU4/PXOxqbrZI9u1i96+7rfBiwQnH7O2+NxmTS7GZFhFBQwfHpzwKxtKS2Hr1vA2eJ8IjB7t7iJlmh5rcjGmCQkE4LHHok8i1rJlZmKoro6ezCFYs7dpgnOPJXRjsiDWJGLffuuei4uzHaFNE5yLLKEbkyWh7e9+m7u/fvr0zN3EIxk7dgQvdioqCt7so6QkeMNtf73V6LPHEroxTVDoFaxjxrhk6cvmFAUQHHtfXR1+O8DQCcj8ETZ28VNmWUI3pombPBl27QqOlPGnCPBv6NEU5qKJ5LfDX3KJS/B+082oUeFJ3RJ+allCNyaHRZuLpimrqXHt8f5wyciE7zfrdOwYPr2BDa1Mjg1bNCYPXXUVTJ3qEqVPJHw5lxXyNAc2bNGYAjN5cvjskd26uWW/qWbMmKbVRFNfNTVuLH/opGUdO7pHotp8PjfzWA3dmAIVOfHYoEFufHxTmpsmFUIvlIo2/06u1fathm6MqSNy2OTkyXXHxp9+em7X5CHYQVtS4troI7+wamrc+sjaei7W5K2GboyJK7Qm36GDW1dd7YZS1ta6aQS+/jp7UwlnSlOpyVsN3RjTYNEmIFMNDqWMdTOQfOPX5ENnsIxWi6+oCLblh47YyQSroRtjUsKvyVdVBWvv3bq5edkBxo51Nft8UFIC27bV7z3NmrkvRf+cNLSmbze4MMY0KRUV4Qm+tBR++EOYPdt9IeTTEMtoGjOjZaOaXETkERFZJyLLYmwfICKbRWSJ97ir/iEaYwpJ5DzxGza45OZfJBXrath8oequE0h1U0wybeiPAmcmKPOmqpZ5j3saH5YxxjihV8P6yd0fhTNmTGbvCJVKqu5XSiolTOiqOh/YmNrDGmNM/UUbahla0/eTfq500FZXp7aWnqpRLv1E5D0ReVFEesQqJCKjRKRSRCrXr1+fokMbY0xQsvd6jTXnfPPmmY339ttTt6+kOkVFpDvwvKoeHWVbe2C3qm4TkUHA/ap6aKJ9WqeoMSabonXM3n+/ez1yZOw7OqWaSP3G8Kd1HLqqblHVbd7r2UCxiHRs7H6NMSadonXM+jf2fuSR+rXVFxdD27YNi+PAAxv2vmgandBFZD8Rd3GwiBzn7TNPRpsaYwpRorb6yM7Z6dPduPT6jshp0yY4Tj8VEja5iMj/AQOAjsBXwN1AMYCqThWRa4AxwC7gG+AGVV2Q6MDW5GKMyVexJgEbPtyNtfcnRGvIBUbxmlwSNv+r6sUJtv8R+GP9QjLGmPzlJ+nQ2Swbc3VosjLcn2uMMYXBb4/PJJucyxhj8oQldGOMyROW0I0xJk9YQjfGmDxhCd0YY/JE1uZDF5H1QFUD394R2JDCcFLF4qofi6t+mmpc0HRjy8e4uqlqp2gbspbQG0NEKmMNrM8mi6t+LK76aapxQdONrdDisiYXY4zJE5bQjTEmT+RqQp+W7QBisLjqx+Kqn6YaFzTd2AoqrpxsQzfGGFNXrtbQjTHGRLCEbowxeSKnErqInCkiH4nIShG5JcuxrBKR90VkiYhUeus6iMirIvKx97xPhmJ5RETWiciykHVRYxFnkncOl4pI7wzHNV5EPvfO2xLvtoX+tlu9uD4Skf9JY1xdRWSeiHwgIstFZKy3PqvnLE5cWT1nItJKRBZ69w1eLiI/89YfJCLveMd/UkRaeOtbessrve3dMxzXoyLyWcj5KvPWZ+zfvne8IhF5V0Se95bTf75UNSceQBHwCXAw0AJ4Dzgqi/GsAjpGrPs1cIv3+hbgfzMUyylAb2BZoliAQcCLgAAnAO9kOK7xwE+jlD3K+5u2BA7y/tZFaYprf6C397od8G/v+Fk9Z3Hiyuo58z53ife6GHjHOw9PARd566cCY7zXVwFTvdcXAU+m6XzFiutRYGiU8hn7t+8d7wbgz7j7MZOJ85VLNfTjgJWq+qmq7gBmAmdnOaZIZwOPea8fA87JxEFVdT6wMclYzgYeV+efwN4isn8G44rlbGCmqm5X1c+Albi/eTriWquqi73XW4EPgc5k+ZzFiSuWjJwz73Nv8xaLvYcCpwFPe+sjz5d/Hp8GThdxt6nMUFyxZOzfvoh0Ac4CHvKWhQycr1xK6J2B1SHLa4j/jz3dFHhFRBaJyChv3b6qutZ7/SWwb3ZCixtLUziP13g/eR8JaZbKSlzez9teuNpdkzlnEXFBls+Z13ywBFgHvIr7NbBJVXdFOfaeuLztm4E4t1hOXVyq6p+vCd75+r2ItIyMK0rMqTYRuAnY7S2XkoHzlUsJvak5SVV7AwOBq0XklNCN6n4/NYkxoU0pFmAK8F2gDFgL/DZbgYhICfAMcL2qbgndls1zFiWurJ8zVa1V1TKgC+5XwBGZjiGayLhE5GjgVlx8fYEOwM2ZjElEBgPrVHVRJo8LuZXQPwe6hix38dZlhap+7j2vA/6G+0f+lf8Tzntel6344sSS1fOoql95/wl3Aw8SbCLIaFwiUoxLmhWq+ldvddbPWbS4mso582LZBMwD+uGaLPzbWIYee09c3va9gOoMxXWm13SlqrodmE7mz1d/YIiIrMI1DZ8G3E8GzlcuJfR/AYd6PcUtcJ0Hz2YjEBFpKyLt/NfA/wOWefEM94oNB/6ejfg8sWJ5Fhjm9fifAGwOaWZIu4g2y3Nx582P6yKvx/8g4FBgYZpiEOBh4ENV/V3Ipqyes1hxZfuciUgnEdnbe90a+D6ufX8eMNQrFnm+/PM4FJjr/eLJRFwrQr6UBddOHXq+0v53VNVbVbWLqnbH5am5qhogE+crVT26mXjgeqn/jWu/uz2LcRyMG13wHrDcjwXX7vUa8DEwB+iQoXj+D/dTfCeube7HsWLB9fA/4J3D94HyDMf1hHfcpd4/5P1Dyt/uxfURMDCNcZ2Ea05ZCizxHoOyfc7ixJXVcwb0BN71jr8MuCvk/8FCXGfsX4CW3vpW3vJKb/vBGY5rrne+lgEzCI6Eydi//ZAYBxAc5ZL282WX/htjTJ7IpSYXY4wxcVhCN8aYPGEJ3Rhj8oQldGOMyROW0I0xJk9YQjfGmDxhCd0YY/LE/wed6Vf/wug9VwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4I6b1u1HEqqM",
        "outputId": "2117ed28-7297-4d22-d67e-32b4870d3eea"
      },
      "source": [
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 2s 7ms/step - loss: 2.2111 - acc: 0.4928\n",
            "accuracy on test set: acc of 49.27999973297119\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2naoRDnf0Hkw"
      },
      "source": [
        "## 3.4 Batch norm model 1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZVBV7rIn12Ui",
        "outputId": "c389a29d-4e66-4c4e-f60c-fb0ae9c0b1fc"
      },
      "source": [
        "#Model with Batch norm type 1\n",
        "from tensorflow.keras.layers import BatchNormalization\n",
        "from tensorflow.keras.initializers import RandomNormal, Constant\n",
        "model = Sequential()\n",
        "# Stack 1\n",
        "model.add(Conv2D(128, (3,3), activation='relu', input_shape=(32, 32, 3)))\n",
        "model.add(Conv2D(128, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "# Stack 2\n",
        "model.add(Conv2D(256, (3,3), activation='relu'))\n",
        "model.add(Conv2D(256, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "# Add a classifier on top of CNN\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 30, 30, 128)       3584      \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 28, 28, 128)       147584    \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 14, 14, 128)       0         \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 14, 14, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 12, 12, 256)       295168    \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 10, 10, 256)       590080    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 5, 5, 256)         0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 5, 5, 256)         0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 6400)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               3277312   \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "batch_normalization (BatchNo (None, 512)               2048      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 100)               51300     \n",
            "=================================================================\n",
            "Total params: 4,367,076\n",
            "Trainable params: 4,366,052\n",
            "Non-trainable params: 1,024\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qN_cOp361cdx"
      },
      "source": [
        "# Data Augmentation\n",
        "# Adding data augmentation for creating more images\n",
        "# Divide train and validation set \n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# Configuration for creating new images\n",
        "train_datagen = ImageDataGenerator(\n",
        "    rotation_range=20,\n",
        "    horizontal_flip=True,\n",
        ")\n",
        "\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(train_images, train_labels, test_size=0.2, random_state=93)\n",
        "train_datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gq9aSzjI1Usg"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_datagen.flow(X_train, y_train, batch_size=64),\n",
        "          steps_per_epoch=100, \n",
        "          epochs=700,\n",
        "          validation_data=(X_validation, y_validation),\n",
        "          verbose=1)\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "0yfzMYjxJ0YG",
        "outputId": "3ae2f314-c0bf-421e-be30-cd28707d4715"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BjvX64idLKWP",
        "outputId": "6b9753a4-67ce-49a7-cd73-58187f6c5d1f"
      },
      "source": [
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 2s 8ms/step - loss: 1.9661 - acc: 0.5448\n",
            "accuracy on test set: acc of 54.47999835014343\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fMWpEqIy6fjN"
      },
      "source": [
        "## 3.5 Batch norm model 2"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ByNLeTCj6kJS",
        "outputId": "937100e5-6965-4f9a-b9ba-a142a81b7e65"
      },
      "source": [
        "# import BatchNormalization\n",
        "from keras.models import Sequential\n",
        "from keras.layers.convolutional import Conv2D\n",
        "from keras.layers.pooling import MaxPool2D\n",
        "from keras.layers.core import Dense,Activation,Dropout,Flatten\n",
        "\n",
        "from tensorflow.keras.layers import BatchNormalization\n",
        "from tensorflow.keras.initializers import RandomNormal, Constant\n",
        "model = Sequential()\n",
        " \n",
        "model.add(Conv2D(128,(3,3),padding='same',input_shape=(32,32,3)))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(128,(3,3),padding='same'))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        " \n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        " \n",
        "model.add(Flatten())\n",
        "model.add(Dense(512))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "model.add(Dense(100,activation='softmax'))\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_4 (Conv2D)            (None, 32, 32, 128)       3584      \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 32, 32, 128)       512       \n",
            "_________________________________________________________________\n",
            "activation (Activation)      (None, 32, 32, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 32, 32, 128)       147584    \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 32, 32, 128)       512       \n",
            "_________________________________________________________________\n",
            "activation_1 (Activation)    (None, 32, 32, 128)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 16, 16, 128)       0         \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 16, 16, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_6 (Conv2D)            (None, 16, 16, 256)       295168    \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 16, 16, 256)       1024      \n",
            "_________________________________________________________________\n",
            "activation_2 (Activation)    (None, 16, 16, 256)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 16, 16, 256)       590080    \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 16, 16, 256)       1024      \n",
            "_________________________________________________________________\n",
            "activation_3 (Activation)    (None, 16, 16, 256)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 8, 8, 256)         0         \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 8, 8, 256)         0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 16384)             0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 512)               8389120   \n",
            "_________________________________________________________________\n",
            "activation_4 (Activation)    (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 100)               51300     \n",
            "=================================================================\n",
            "Total params: 9,479,908\n",
            "Trainable params: 9,478,372\n",
            "Non-trainable params: 1,536\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mzk68oXN6umT"
      },
      "source": [
        "# Data Augmentation\n",
        "# Adding data augmentation for creating more images\n",
        "# Divide train and validation set \n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# Configuration for creating new images\n",
        "train_datagen = ImageDataGenerator(\n",
        "    rotation_range=20,\n",
        "    horizontal_flip=True,\n",
        ")\n",
        "\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(train_images, train_labels, test_size=0.2, random_state=93)\n",
        "train_datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ckqlQTYi6w75"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_datagen.flow(X_train, y_train, batch_size=64),\n",
        "          steps_per_epoch=100, \n",
        "          epochs=700,\n",
        "          validation_data=(X_validation, y_validation),\n",
        "          verbose=1)\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        },
        "id": "_xVUXZSgMj9u",
        "outputId": "4017c996-a053-4267-8d77-95d1e9488730"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R6pSux277Wup"
      },
      "source": [
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "v0VXUeE_1kjW"
      },
      "source": [
        "## 3.6 Batch Norm model 3"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LfDHil2r-kLw",
        "outputId": "6ea64584-e796-46ba-9a40-50b2b55b14b0"
      },
      "source": [
        "# Batch norm model 3\n",
        "from keras.models import Sequential\n",
        "from keras.layers.convolutional import Conv2D\n",
        "from keras.layers.pooling import MaxPool2D\n",
        "from keras.layers.core import Dense,Activation,Dropout,Flatten\n",
        "\n",
        "from tensorflow.keras.layers import BatchNormalization\n",
        "from tensorflow.keras.initializers import RandomNormal, Constant\n",
        "model = Sequential()\n",
        " \n",
        "model.add(Conv2D(128,(3,3),padding='same',input_shape=(32,32,3)))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(128,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        " \n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Dense(100,activation='softmax'))\n",
        "model.summary()\n",
        "\n",
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_10 (Conv2D)           (None, 32, 32, 128)       3584      \n",
            "_________________________________________________________________\n",
            "batch_normalization_12 (Batc (None, 32, 32, 128)       512       \n",
            "_________________________________________________________________\n",
            "activation_12 (Activation)   (None, 32, 32, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_11 (Conv2D)           (None, 32, 32, 128)       147584    \n",
            "_________________________________________________________________\n",
            "batch_normalization_13 (Batc (None, 32, 32, 128)       512       \n",
            "_________________________________________________________________\n",
            "activation_13 (Activation)   (None, 32, 32, 128)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_5 (MaxPooling2 (None, 16, 16, 128)       0         \n",
            "_________________________________________________________________\n",
            "dropout_7 (Dropout)          (None, 16, 16, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_12 (Conv2D)           (None, 16, 16, 256)       295168    \n",
            "_________________________________________________________________\n",
            "batch_normalization_14 (Batc (None, 16, 16, 256)       1024      \n",
            "_________________________________________________________________\n",
            "activation_14 (Activation)   (None, 16, 16, 256)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_13 (Conv2D)           (None, 16, 16, 256)       590080    \n",
            "_________________________________________________________________\n",
            "batch_normalization_15 (Batc (None, 16, 16, 256)       1024      \n",
            "_________________________________________________________________\n",
            "activation_15 (Activation)   (None, 16, 16, 256)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_6 (MaxPooling2 (None, 8, 8, 256)         0         \n",
            "_________________________________________________________________\n",
            "dropout_8 (Dropout)          (None, 8, 8, 256)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_14 (Conv2D)           (None, 8, 8, 256)         590080    \n",
            "_________________________________________________________________\n",
            "batch_normalization_16 (Batc (None, 8, 8, 256)         1024      \n",
            "_________________________________________________________________\n",
            "activation_16 (Activation)   (None, 8, 8, 256)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_15 (Conv2D)           (None, 8, 8, 256)         590080    \n",
            "_________________________________________________________________\n",
            "batch_normalization_17 (Batc (None, 8, 8, 256)         1024      \n",
            "_________________________________________________________________\n",
            "activation_17 (Activation)   (None, 8, 8, 256)         0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_7 (MaxPooling2 (None, 4, 4, 256)         0         \n",
            "_________________________________________________________________\n",
            "dropout_9 (Dropout)          (None, 4, 4, 256)         0         \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 4096)              0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 512)               2097664   \n",
            "_________________________________________________________________\n",
            "activation_18 (Activation)   (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dropout_10 (Dropout)         (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_18 (Batc (None, 512)               2048      \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 100)               51300     \n",
            "=================================================================\n",
            "Total params: 4,372,708\n",
            "Trainable params: 4,369,124\n",
            "Non-trainable params: 3,584\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ds2FPVLE-tuM"
      },
      "source": [
        "# Data Augmentation\n",
        "# Adding data augmentation for creating more images\n",
        "# Divide train and validation set \n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# Configuration for creating new images\n",
        "train_datagen = ImageDataGenerator(\n",
        "    rotation_range=20,\n",
        "    horizontal_flip=True,\n",
        ")\n",
        "\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(train_images, train_labels, test_size=0.2, random_state=93)\n",
        "train_datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sX38v6Dz-053",
        "outputId": "6263c484-279a-4404-b325-d61735f86438"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_datagen.flow(X_train, y_train, batch_size=64),\n",
        "          steps_per_epoch=100, \n",
        "          epochs=350,\n",
        "          validation_data=(X_validation, y_validation),\n",
        "          verbose=1)\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "100/100 [==============================] - 12s 116ms/step - loss: 1.9742 - acc: 0.4778 - val_loss: 2.0093 - val_acc: 0.4676\n",
            "Epoch 59/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 2.0124 - acc: 0.4719 - val_loss: 2.2211 - val_acc: 0.4331\n",
            "Epoch 60/350\n",
            "100/100 [==============================] - 11s 115ms/step - loss: 1.9399 - acc: 0.4811 - val_loss: 2.1851 - val_acc: 0.4337\n",
            "Epoch 61/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.9640 - acc: 0.4858 - val_loss: 2.0321 - val_acc: 0.4692\n",
            "Epoch 62/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.9379 - acc: 0.4847 - val_loss: 1.9887 - val_acc: 0.4752\n",
            "Epoch 63/350\n",
            "100/100 [==============================] - 13s 126ms/step - loss: 1.9462 - acc: 0.4848 - val_loss: 2.1070 - val_acc: 0.4450\n",
            "Epoch 64/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.9090 - acc: 0.4914 - val_loss: 1.9928 - val_acc: 0.4746\n",
            "Epoch 65/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.8911 - acc: 0.4977 - val_loss: 2.0372 - val_acc: 0.4663\n",
            "Epoch 66/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.8425 - acc: 0.5106 - val_loss: 2.1744 - val_acc: 0.4387\n",
            "Epoch 67/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.8643 - acc: 0.4992 - val_loss: 2.0453 - val_acc: 0.4640\n",
            "Epoch 68/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.8866 - acc: 0.4984 - val_loss: 2.0278 - val_acc: 0.4683\n",
            "Epoch 69/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.8293 - acc: 0.5108 - val_loss: 1.9895 - val_acc: 0.4794\n",
            "Epoch 70/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.8616 - acc: 0.5072 - val_loss: 2.0035 - val_acc: 0.4723\n",
            "Epoch 71/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.8411 - acc: 0.5078 - val_loss: 1.9558 - val_acc: 0.4800\n",
            "Epoch 72/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7927 - acc: 0.5336 - val_loss: 1.9675 - val_acc: 0.4799\n",
            "Epoch 73/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.8053 - acc: 0.5192 - val_loss: 1.9426 - val_acc: 0.4851\n",
            "Epoch 74/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.7946 - acc: 0.5228 - val_loss: 1.9748 - val_acc: 0.4716\n",
            "Epoch 75/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7740 - acc: 0.5286 - val_loss: 2.0223 - val_acc: 0.4702\n",
            "Epoch 76/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.7849 - acc: 0.5238 - val_loss: 1.9334 - val_acc: 0.4866\n",
            "Epoch 77/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7875 - acc: 0.5131 - val_loss: 1.8654 - val_acc: 0.4996\n",
            "Epoch 78/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7553 - acc: 0.5245 - val_loss: 1.8719 - val_acc: 0.5003\n",
            "Epoch 79/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7637 - acc: 0.5238 - val_loss: 1.8974 - val_acc: 0.4969\n",
            "Epoch 80/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7096 - acc: 0.5419 - val_loss: 1.7720 - val_acc: 0.5202\n",
            "Epoch 81/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7422 - acc: 0.5350 - val_loss: 1.9679 - val_acc: 0.4803\n",
            "Epoch 82/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.7214 - acc: 0.5352 - val_loss: 1.8883 - val_acc: 0.4907\n",
            "Epoch 83/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.6977 - acc: 0.5367 - val_loss: 1.8920 - val_acc: 0.4932\n",
            "Epoch 84/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.6971 - acc: 0.5391 - val_loss: 1.9351 - val_acc: 0.4907\n",
            "Epoch 85/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.7010 - acc: 0.5417 - val_loss: 1.8664 - val_acc: 0.5021\n",
            "Epoch 86/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6630 - acc: 0.5505 - val_loss: 2.0061 - val_acc: 0.4772\n",
            "Epoch 87/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.6412 - acc: 0.5511 - val_loss: 2.0235 - val_acc: 0.4802\n",
            "Epoch 88/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6479 - acc: 0.5511 - val_loss: 1.7979 - val_acc: 0.5170\n",
            "Epoch 89/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6514 - acc: 0.5602 - val_loss: 1.8161 - val_acc: 0.5174\n",
            "Epoch 90/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6731 - acc: 0.5534 - val_loss: 1.9920 - val_acc: 0.4810\n",
            "Epoch 91/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6117 - acc: 0.5680 - val_loss: 1.8332 - val_acc: 0.5072\n",
            "Epoch 92/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6556 - acc: 0.5566 - val_loss: 1.8085 - val_acc: 0.5138\n",
            "Epoch 93/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.6187 - acc: 0.5661 - val_loss: 1.7720 - val_acc: 0.5260\n",
            "Epoch 94/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.6005 - acc: 0.5677 - val_loss: 1.7561 - val_acc: 0.5300\n",
            "Epoch 95/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.6146 - acc: 0.5702 - val_loss: 1.7623 - val_acc: 0.5229\n",
            "Epoch 96/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5661 - acc: 0.5788 - val_loss: 1.8254 - val_acc: 0.5138\n",
            "Epoch 97/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5783 - acc: 0.5692 - val_loss: 1.7859 - val_acc: 0.5177\n",
            "Epoch 98/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5532 - acc: 0.5783 - val_loss: 1.7983 - val_acc: 0.5166\n",
            "Epoch 99/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5419 - acc: 0.5783 - val_loss: 1.7588 - val_acc: 0.5208\n",
            "Epoch 100/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.5567 - acc: 0.5739 - val_loss: 1.7861 - val_acc: 0.5188\n",
            "Epoch 101/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.5653 - acc: 0.5808 - val_loss: 1.7552 - val_acc: 0.5270\n",
            "Epoch 102/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5224 - acc: 0.5891 - val_loss: 1.8296 - val_acc: 0.5116\n",
            "Epoch 103/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5460 - acc: 0.5758 - val_loss: 1.7723 - val_acc: 0.5257\n",
            "Epoch 104/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.5438 - acc: 0.5808 - val_loss: 1.7909 - val_acc: 0.5213\n",
            "Epoch 105/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.5226 - acc: 0.5814 - val_loss: 1.8085 - val_acc: 0.5150\n",
            "Epoch 106/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.5091 - acc: 0.5944 - val_loss: 1.7443 - val_acc: 0.5332\n",
            "Epoch 107/350\n",
            "100/100 [==============================] - 11s 115ms/step - loss: 1.5284 - acc: 0.5841 - val_loss: 1.6917 - val_acc: 0.5416\n",
            "Epoch 108/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.5235 - acc: 0.5800 - val_loss: 1.6756 - val_acc: 0.5496\n",
            "Epoch 109/350\n",
            "100/100 [==============================] - 11s 115ms/step - loss: 1.4980 - acc: 0.5923 - val_loss: 1.8005 - val_acc: 0.5191\n",
            "Epoch 110/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4877 - acc: 0.5864 - val_loss: 1.7660 - val_acc: 0.5298\n",
            "Epoch 111/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4586 - acc: 0.6006 - val_loss: 1.6511 - val_acc: 0.5533\n",
            "Epoch 112/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.4703 - acc: 0.5934 - val_loss: 1.8210 - val_acc: 0.5178\n",
            "Epoch 113/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4640 - acc: 0.6014 - val_loss: 1.7452 - val_acc: 0.5234\n",
            "Epoch 114/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4504 - acc: 0.5980 - val_loss: 1.6621 - val_acc: 0.5511\n",
            "Epoch 115/350\n",
            "100/100 [==============================] - 13s 126ms/step - loss: 1.4787 - acc: 0.5934 - val_loss: 1.8083 - val_acc: 0.5167\n",
            "Epoch 116/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4650 - acc: 0.5989 - val_loss: 1.6788 - val_acc: 0.5466\n",
            "Epoch 117/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.4412 - acc: 0.6036 - val_loss: 1.7184 - val_acc: 0.5413\n",
            "Epoch 118/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4217 - acc: 0.6081 - val_loss: 1.8125 - val_acc: 0.5125\n",
            "Epoch 119/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.4433 - acc: 0.6055 - val_loss: 1.8543 - val_acc: 0.5065\n",
            "Epoch 120/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.3966 - acc: 0.6187 - val_loss: 1.6502 - val_acc: 0.5551\n",
            "Epoch 121/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.4226 - acc: 0.6048 - val_loss: 1.7266 - val_acc: 0.5312\n",
            "Epoch 122/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.4097 - acc: 0.6041 - val_loss: 1.6847 - val_acc: 0.5479\n",
            "Epoch 123/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.3935 - acc: 0.6161 - val_loss: 1.8003 - val_acc: 0.5229\n",
            "Epoch 124/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.3966 - acc: 0.6102 - val_loss: 1.8148 - val_acc: 0.5194\n",
            "Epoch 125/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3663 - acc: 0.6267 - val_loss: 1.6266 - val_acc: 0.5594\n",
            "Epoch 126/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.3589 - acc: 0.6197 - val_loss: 1.6138 - val_acc: 0.5611\n",
            "Epoch 127/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.3448 - acc: 0.6264 - val_loss: 1.7534 - val_acc: 0.5280\n",
            "Epoch 128/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.3732 - acc: 0.6189 - val_loss: 1.6313 - val_acc: 0.5607\n",
            "Epoch 129/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.3469 - acc: 0.6236 - val_loss: 1.5970 - val_acc: 0.5624\n",
            "Epoch 130/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3363 - acc: 0.6248 - val_loss: 1.7804 - val_acc: 0.5265\n",
            "Epoch 131/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3513 - acc: 0.6255 - val_loss: 1.6047 - val_acc: 0.5653\n",
            "Epoch 132/350\n",
            "100/100 [==============================] - 13s 127ms/step - loss: 1.3419 - acc: 0.6277 - val_loss: 1.5708 - val_acc: 0.5720\n",
            "Epoch 133/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3581 - acc: 0.6194 - val_loss: 1.6857 - val_acc: 0.5439\n",
            "Epoch 134/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.3301 - acc: 0.6331 - val_loss: 1.7273 - val_acc: 0.5402\n",
            "Epoch 135/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3197 - acc: 0.6333 - val_loss: 1.6989 - val_acc: 0.5411\n",
            "Epoch 136/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.3318 - acc: 0.6336 - val_loss: 1.5717 - val_acc: 0.5738\n",
            "Epoch 137/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3034 - acc: 0.6361 - val_loss: 1.7476 - val_acc: 0.5326\n",
            "Epoch 138/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2811 - acc: 0.6467 - val_loss: 1.9034 - val_acc: 0.5078\n",
            "Epoch 139/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2857 - acc: 0.6444 - val_loss: 1.6943 - val_acc: 0.5447\n",
            "Epoch 140/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3109 - acc: 0.6356 - val_loss: 1.6327 - val_acc: 0.5605\n",
            "Epoch 141/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2645 - acc: 0.6505 - val_loss: 1.5985 - val_acc: 0.5639\n",
            "Epoch 142/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.3116 - acc: 0.6328 - val_loss: 1.6501 - val_acc: 0.5561\n",
            "Epoch 143/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.2899 - acc: 0.6420 - val_loss: 1.7163 - val_acc: 0.5453\n",
            "Epoch 144/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2882 - acc: 0.6470 - val_loss: 1.6184 - val_acc: 0.5674\n",
            "Epoch 145/350\n",
            "100/100 [==============================] - 13s 127ms/step - loss: 1.2721 - acc: 0.6488 - val_loss: 1.6586 - val_acc: 0.5547\n",
            "Epoch 146/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2459 - acc: 0.6498 - val_loss: 1.6219 - val_acc: 0.5655\n",
            "Epoch 147/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2540 - acc: 0.6497 - val_loss: 1.7194 - val_acc: 0.5373\n",
            "Epoch 148/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2545 - acc: 0.6506 - val_loss: 1.6512 - val_acc: 0.5553\n",
            "Epoch 149/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.2524 - acc: 0.6556 - val_loss: 1.6240 - val_acc: 0.5626\n",
            "Epoch 150/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2348 - acc: 0.6550 - val_loss: 1.8239 - val_acc: 0.5223\n",
            "Epoch 151/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2201 - acc: 0.6603 - val_loss: 1.7197 - val_acc: 0.5420\n",
            "Epoch 152/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.2426 - acc: 0.6506 - val_loss: 1.6454 - val_acc: 0.5549\n",
            "Epoch 153/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2148 - acc: 0.6597 - val_loss: 1.6163 - val_acc: 0.5645\n",
            "Epoch 154/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.2319 - acc: 0.6525 - val_loss: 1.6982 - val_acc: 0.5432\n",
            "Epoch 155/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2166 - acc: 0.6586 - val_loss: 1.6188 - val_acc: 0.5628\n",
            "Epoch 156/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2184 - acc: 0.6577 - val_loss: 1.5787 - val_acc: 0.5719\n",
            "Epoch 157/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.2311 - acc: 0.6564 - val_loss: 1.6136 - val_acc: 0.5716\n",
            "Epoch 158/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1987 - acc: 0.6669 - val_loss: 1.5885 - val_acc: 0.5723\n",
            "Epoch 159/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1839 - acc: 0.6662 - val_loss: 1.6766 - val_acc: 0.5472\n",
            "Epoch 160/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1972 - acc: 0.6634 - val_loss: 1.6486 - val_acc: 0.5549\n",
            "Epoch 161/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1910 - acc: 0.6558 - val_loss: 1.6230 - val_acc: 0.5687\n",
            "Epoch 162/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1885 - acc: 0.6667 - val_loss: 1.5885 - val_acc: 0.5726\n",
            "Epoch 163/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1488 - acc: 0.6737 - val_loss: 1.5471 - val_acc: 0.5788\n",
            "Epoch 164/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1967 - acc: 0.6595 - val_loss: 1.6351 - val_acc: 0.5589\n",
            "Epoch 165/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1663 - acc: 0.6750 - val_loss: 1.7101 - val_acc: 0.5460\n",
            "Epoch 166/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1938 - acc: 0.6637 - val_loss: 1.6191 - val_acc: 0.5586\n",
            "Epoch 167/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1993 - acc: 0.6609 - val_loss: 1.6189 - val_acc: 0.5629\n",
            "Epoch 168/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1869 - acc: 0.6689 - val_loss: 1.6127 - val_acc: 0.5661\n",
            "Epoch 169/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.1602 - acc: 0.6739 - val_loss: 1.5233 - val_acc: 0.5832\n",
            "Epoch 170/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1605 - acc: 0.6725 - val_loss: 1.5585 - val_acc: 0.5830\n",
            "Epoch 171/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1278 - acc: 0.6808 - val_loss: 1.5408 - val_acc: 0.5867\n",
            "Epoch 172/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1547 - acc: 0.6739 - val_loss: 1.5277 - val_acc: 0.5816\n",
            "Epoch 173/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1624 - acc: 0.6747 - val_loss: 1.6019 - val_acc: 0.5694\n",
            "Epoch 174/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1398 - acc: 0.6741 - val_loss: 1.6056 - val_acc: 0.5657\n",
            "Epoch 175/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.0993 - acc: 0.6891 - val_loss: 1.5871 - val_acc: 0.5727\n",
            "Epoch 176/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.1069 - acc: 0.6861 - val_loss: 1.5368 - val_acc: 0.5899\n",
            "Epoch 177/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1243 - acc: 0.6820 - val_loss: 1.5612 - val_acc: 0.5802\n",
            "Epoch 178/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1192 - acc: 0.6816 - val_loss: 1.5364 - val_acc: 0.5836\n",
            "Epoch 179/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.1151 - acc: 0.6852 - val_loss: 1.6712 - val_acc: 0.5548\n",
            "Epoch 180/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 1.0959 - acc: 0.6917 - val_loss: 1.5630 - val_acc: 0.5772\n",
            "Epoch 181/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 1.1253 - acc: 0.6830 - val_loss: 1.5420 - val_acc: 0.5808\n",
            "Epoch 182/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1201 - acc: 0.6823 - val_loss: 1.6432 - val_acc: 0.5633\n",
            "Epoch 183/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0852 - acc: 0.6862 - val_loss: 1.5530 - val_acc: 0.5787\n",
            "Epoch 184/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.1250 - acc: 0.6811 - val_loss: 1.5251 - val_acc: 0.5812\n",
            "Epoch 185/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0963 - acc: 0.6916 - val_loss: 1.5648 - val_acc: 0.5796\n",
            "Epoch 186/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0758 - acc: 0.6950 - val_loss: 1.6599 - val_acc: 0.5577\n",
            "Epoch 187/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.1029 - acc: 0.6886 - val_loss: 1.5564 - val_acc: 0.5799\n",
            "Epoch 188/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0465 - acc: 0.7059 - val_loss: 1.5348 - val_acc: 0.5857\n",
            "Epoch 189/350\n",
            "100/100 [==============================] - 13s 128ms/step - loss: 1.0686 - acc: 0.6948 - val_loss: 1.5754 - val_acc: 0.5767\n",
            "Epoch 190/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0745 - acc: 0.6983 - val_loss: 1.6646 - val_acc: 0.5578\n",
            "Epoch 191/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0593 - acc: 0.6986 - val_loss: 1.5373 - val_acc: 0.5852\n",
            "Epoch 192/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 1.0750 - acc: 0.6975 - val_loss: 1.5307 - val_acc: 0.5853\n",
            "Epoch 193/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0569 - acc: 0.7072 - val_loss: 1.4942 - val_acc: 0.5967\n",
            "Epoch 194/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0699 - acc: 0.6978 - val_loss: 1.5522 - val_acc: 0.5838\n",
            "Epoch 195/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 1.0739 - acc: 0.6913 - val_loss: 1.4868 - val_acc: 0.6019\n",
            "Epoch 196/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0552 - acc: 0.6963 - val_loss: 1.5475 - val_acc: 0.5838\n",
            "Epoch 197/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0282 - acc: 0.7123 - val_loss: 1.6474 - val_acc: 0.5693\n",
            "Epoch 198/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0355 - acc: 0.7013 - val_loss: 1.5354 - val_acc: 0.5864\n",
            "Epoch 199/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0163 - acc: 0.7105 - val_loss: 1.5310 - val_acc: 0.5904\n",
            "Epoch 200/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0508 - acc: 0.7039 - val_loss: 1.5801 - val_acc: 0.5730\n",
            "Epoch 201/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0252 - acc: 0.7088 - val_loss: 1.5019 - val_acc: 0.5939\n",
            "Epoch 202/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0164 - acc: 0.7127 - val_loss: 1.5763 - val_acc: 0.5770\n",
            "Epoch 203/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0492 - acc: 0.7031 - val_loss: 1.5187 - val_acc: 0.5922\n",
            "Epoch 204/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0135 - acc: 0.7111 - val_loss: 1.5058 - val_acc: 0.5924\n",
            "Epoch 205/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0464 - acc: 0.7003 - val_loss: 1.4717 - val_acc: 0.6014\n",
            "Epoch 206/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0379 - acc: 0.7014 - val_loss: 1.5215 - val_acc: 0.5931\n",
            "Epoch 207/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9982 - acc: 0.7119 - val_loss: 1.5250 - val_acc: 0.5908\n",
            "Epoch 208/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 0.9976 - acc: 0.7142 - val_loss: 1.4999 - val_acc: 0.6031\n",
            "Epoch 209/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 1.0085 - acc: 0.7127 - val_loss: 1.4927 - val_acc: 0.5976\n",
            "Epoch 210/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9617 - acc: 0.7242 - val_loss: 1.4953 - val_acc: 0.5985\n",
            "Epoch 211/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 0.9886 - acc: 0.7178 - val_loss: 1.4778 - val_acc: 0.5966\n",
            "Epoch 212/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 1.0151 - acc: 0.7064 - val_loss: 1.4656 - val_acc: 0.6019\n",
            "Epoch 213/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9726 - acc: 0.7159 - val_loss: 1.4710 - val_acc: 0.6002\n",
            "Epoch 214/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9734 - acc: 0.7241 - val_loss: 1.4802 - val_acc: 0.6053\n",
            "Epoch 215/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9820 - acc: 0.7147 - val_loss: 1.4815 - val_acc: 0.6029\n",
            "Epoch 216/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9865 - acc: 0.7150 - val_loss: 1.5771 - val_acc: 0.5759\n",
            "Epoch 217/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9599 - acc: 0.7266 - val_loss: 1.5720 - val_acc: 0.5760\n",
            "Epoch 218/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.9928 - acc: 0.7169 - val_loss: 1.4219 - val_acc: 0.6124\n",
            "Epoch 219/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9680 - acc: 0.7184 - val_loss: 1.5172 - val_acc: 0.5952\n",
            "Epoch 220/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9596 - acc: 0.7278 - val_loss: 1.5129 - val_acc: 0.5950\n",
            "Epoch 221/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9562 - acc: 0.7250 - val_loss: 1.4995 - val_acc: 0.5980\n",
            "Epoch 222/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9475 - acc: 0.7250 - val_loss: 1.4719 - val_acc: 0.6081\n",
            "Epoch 223/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9339 - acc: 0.7314 - val_loss: 1.5553 - val_acc: 0.5867\n",
            "Epoch 224/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9442 - acc: 0.7336 - val_loss: 1.4737 - val_acc: 0.5991\n",
            "Epoch 225/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9926 - acc: 0.7158 - val_loss: 1.5328 - val_acc: 0.5854\n",
            "Epoch 226/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9397 - acc: 0.7270 - val_loss: 1.4849 - val_acc: 0.5965\n",
            "Epoch 227/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9509 - acc: 0.7317 - val_loss: 1.4790 - val_acc: 0.6047\n",
            "Epoch 228/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9404 - acc: 0.7286 - val_loss: 1.6893 - val_acc: 0.5610\n",
            "Epoch 229/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9134 - acc: 0.7347 - val_loss: 1.5394 - val_acc: 0.5854\n",
            "Epoch 230/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9446 - acc: 0.7275 - val_loss: 1.4580 - val_acc: 0.6105\n",
            "Epoch 231/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9064 - acc: 0.7356 - val_loss: 1.4475 - val_acc: 0.6110\n",
            "Epoch 232/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9449 - acc: 0.7252 - val_loss: 1.5230 - val_acc: 0.5908\n",
            "Epoch 233/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9249 - acc: 0.7280 - val_loss: 1.4550 - val_acc: 0.6117\n",
            "Epoch 234/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9219 - acc: 0.7347 - val_loss: 1.5449 - val_acc: 0.5879\n",
            "Epoch 235/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9078 - acc: 0.7402 - val_loss: 1.5236 - val_acc: 0.5897\n",
            "Epoch 236/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8955 - acc: 0.7402 - val_loss: 1.4918 - val_acc: 0.6011\n",
            "Epoch 237/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9240 - acc: 0.7370 - val_loss: 1.5118 - val_acc: 0.5930\n",
            "Epoch 238/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9102 - acc: 0.7375 - val_loss: 1.5610 - val_acc: 0.5896\n",
            "Epoch 239/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9084 - acc: 0.7377 - val_loss: 1.5220 - val_acc: 0.5894\n",
            "Epoch 240/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8972 - acc: 0.7417 - val_loss: 1.5974 - val_acc: 0.5811\n",
            "Epoch 241/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9078 - acc: 0.7350 - val_loss: 1.5088 - val_acc: 0.6006\n",
            "Epoch 242/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.9144 - acc: 0.7372 - val_loss: 1.4673 - val_acc: 0.6098\n",
            "Epoch 243/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9209 - acc: 0.7405 - val_loss: 1.5862 - val_acc: 0.5800\n",
            "Epoch 244/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8894 - acc: 0.7377 - val_loss: 1.5563 - val_acc: 0.5899\n",
            "Epoch 245/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8884 - acc: 0.7484 - val_loss: 1.6409 - val_acc: 0.5747\n",
            "Epoch 246/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8864 - acc: 0.7452 - val_loss: 1.4782 - val_acc: 0.6051\n",
            "Epoch 247/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9192 - acc: 0.7348 - val_loss: 1.4439 - val_acc: 0.6118\n",
            "Epoch 248/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8812 - acc: 0.7467 - val_loss: 1.5034 - val_acc: 0.5977\n",
            "Epoch 249/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8944 - acc: 0.7416 - val_loss: 1.4850 - val_acc: 0.6054\n",
            "Epoch 250/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 0.8751 - acc: 0.7462 - val_loss: 1.4347 - val_acc: 0.6189\n",
            "Epoch 251/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.9108 - acc: 0.7403 - val_loss: 1.5364 - val_acc: 0.5936\n",
            "Epoch 252/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8659 - acc: 0.7505 - val_loss: 1.5106 - val_acc: 0.5988\n",
            "Epoch 253/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8416 - acc: 0.7522 - val_loss: 1.4810 - val_acc: 0.6041\n",
            "Epoch 254/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8858 - acc: 0.7453 - val_loss: 1.4304 - val_acc: 0.6169\n",
            "Epoch 255/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8669 - acc: 0.7516 - val_loss: 1.4369 - val_acc: 0.6159\n",
            "Epoch 256/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8615 - acc: 0.7573 - val_loss: 1.4880 - val_acc: 0.6053\n",
            "Epoch 257/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8661 - acc: 0.7505 - val_loss: 1.4233 - val_acc: 0.6194\n",
            "Epoch 258/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8327 - acc: 0.7581 - val_loss: 1.4744 - val_acc: 0.6062\n",
            "Epoch 259/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8452 - acc: 0.7517 - val_loss: 1.4888 - val_acc: 0.6082\n",
            "Epoch 260/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8358 - acc: 0.7539 - val_loss: 1.4731 - val_acc: 0.6085\n",
            "Epoch 261/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8516 - acc: 0.7575 - val_loss: 1.5060 - val_acc: 0.6044\n",
            "Epoch 262/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8317 - acc: 0.7573 - val_loss: 1.4721 - val_acc: 0.6093\n",
            "Epoch 263/350\n",
            "100/100 [==============================] - 12s 115ms/step - loss: 0.8501 - acc: 0.7539 - val_loss: 1.4904 - val_acc: 0.6062\n",
            "Epoch 264/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8403 - acc: 0.7578 - val_loss: 1.5446 - val_acc: 0.5976\n",
            "Epoch 265/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8198 - acc: 0.7622 - val_loss: 1.4718 - val_acc: 0.6121\n",
            "Epoch 266/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8455 - acc: 0.7458 - val_loss: 1.4598 - val_acc: 0.6118\n",
            "Epoch 267/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8148 - acc: 0.7598 - val_loss: 1.5018 - val_acc: 0.6050\n",
            "Epoch 268/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8146 - acc: 0.7627 - val_loss: 1.4011 - val_acc: 0.6301\n",
            "Epoch 269/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8514 - acc: 0.7467 - val_loss: 1.5286 - val_acc: 0.6041\n",
            "Epoch 270/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8283 - acc: 0.7589 - val_loss: 1.4783 - val_acc: 0.6092\n",
            "Epoch 271/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8220 - acc: 0.7634 - val_loss: 1.5368 - val_acc: 0.5974\n",
            "Epoch 272/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8228 - acc: 0.7641 - val_loss: 1.4084 - val_acc: 0.6225\n",
            "Epoch 273/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8123 - acc: 0.7609 - val_loss: 1.5485 - val_acc: 0.5911\n",
            "Epoch 274/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8388 - acc: 0.7598 - val_loss: 1.5271 - val_acc: 0.5995\n",
            "Epoch 275/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8111 - acc: 0.7625 - val_loss: 1.4642 - val_acc: 0.6109\n",
            "Epoch 276/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7990 - acc: 0.7659 - val_loss: 1.5383 - val_acc: 0.5961\n",
            "Epoch 277/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7905 - acc: 0.7714 - val_loss: 1.5004 - val_acc: 0.6039\n",
            "Epoch 278/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8179 - acc: 0.7630 - val_loss: 1.4252 - val_acc: 0.6152\n",
            "Epoch 279/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7849 - acc: 0.7728 - val_loss: 1.4518 - val_acc: 0.6159\n",
            "Epoch 280/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8030 - acc: 0.7592 - val_loss: 1.4107 - val_acc: 0.6241\n",
            "Epoch 281/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8104 - acc: 0.7614 - val_loss: 1.4651 - val_acc: 0.6140\n",
            "Epoch 282/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8136 - acc: 0.7598 - val_loss: 1.4135 - val_acc: 0.6201\n",
            "Epoch 283/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.8002 - acc: 0.7634 - val_loss: 1.4610 - val_acc: 0.6190\n",
            "Epoch 284/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8055 - acc: 0.7638 - val_loss: 1.4703 - val_acc: 0.6101\n",
            "Epoch 285/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.8087 - acc: 0.7673 - val_loss: 1.4937 - val_acc: 0.6100\n",
            "Epoch 286/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7772 - acc: 0.7750 - val_loss: 1.5054 - val_acc: 0.6105\n",
            "Epoch 287/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7761 - acc: 0.7711 - val_loss: 1.4920 - val_acc: 0.6131\n",
            "Epoch 288/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7789 - acc: 0.7703 - val_loss: 1.5347 - val_acc: 0.5971\n",
            "Epoch 289/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7840 - acc: 0.7673 - val_loss: 1.4613 - val_acc: 0.6190\n",
            "Epoch 290/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7715 - acc: 0.7738 - val_loss: 1.5147 - val_acc: 0.5992\n",
            "Epoch 291/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7332 - acc: 0.7897 - val_loss: 1.4997 - val_acc: 0.6160\n",
            "Epoch 292/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7529 - acc: 0.7761 - val_loss: 1.5143 - val_acc: 0.6029\n",
            "Epoch 293/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7756 - acc: 0.7723 - val_loss: 1.4453 - val_acc: 0.6160\n",
            "Epoch 294/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7600 - acc: 0.7822 - val_loss: 1.4587 - val_acc: 0.6188\n",
            "Epoch 295/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.7643 - acc: 0.7773 - val_loss: 1.4536 - val_acc: 0.6176\n",
            "Epoch 296/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7550 - acc: 0.7825 - val_loss: 1.5698 - val_acc: 0.5932\n",
            "Epoch 297/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7754 - acc: 0.7708 - val_loss: 1.4869 - val_acc: 0.6070\n",
            "Epoch 298/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7646 - acc: 0.7777 - val_loss: 1.5276 - val_acc: 0.6004\n",
            "Epoch 299/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7484 - acc: 0.7802 - val_loss: 1.5194 - val_acc: 0.6075\n",
            "Epoch 300/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7593 - acc: 0.7794 - val_loss: 1.4873 - val_acc: 0.6149\n",
            "Epoch 301/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7485 - acc: 0.7784 - val_loss: 1.4451 - val_acc: 0.6168\n",
            "Epoch 302/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7400 - acc: 0.7781 - val_loss: 1.4259 - val_acc: 0.6240\n",
            "Epoch 303/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7551 - acc: 0.7802 - val_loss: 1.4880 - val_acc: 0.6115\n",
            "Epoch 304/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7292 - acc: 0.7880 - val_loss: 1.4545 - val_acc: 0.6213\n",
            "Epoch 305/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7288 - acc: 0.7903 - val_loss: 1.4558 - val_acc: 0.6213\n",
            "Epoch 306/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7320 - acc: 0.7903 - val_loss: 1.4705 - val_acc: 0.6141\n",
            "Epoch 307/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7474 - acc: 0.7781 - val_loss: 1.4616 - val_acc: 0.6170\n",
            "Epoch 308/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7016 - acc: 0.7969 - val_loss: 1.5049 - val_acc: 0.6099\n",
            "Epoch 309/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7064 - acc: 0.7927 - val_loss: 1.5458 - val_acc: 0.6025\n",
            "Epoch 310/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7168 - acc: 0.7891 - val_loss: 1.4223 - val_acc: 0.6247\n",
            "Epoch 311/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7262 - acc: 0.7898 - val_loss: 1.4791 - val_acc: 0.6131\n",
            "Epoch 312/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7185 - acc: 0.7850 - val_loss: 1.4740 - val_acc: 0.6114\n",
            "Epoch 313/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7187 - acc: 0.7875 - val_loss: 1.5138 - val_acc: 0.6059\n",
            "Epoch 314/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6965 - acc: 0.7984 - val_loss: 1.4069 - val_acc: 0.6262\n",
            "Epoch 315/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.7307 - acc: 0.7859 - val_loss: 1.3847 - val_acc: 0.6356\n",
            "Epoch 316/350\n",
            "100/100 [==============================] - 13s 128ms/step - loss: 0.7075 - acc: 0.7927 - val_loss: 1.7447 - val_acc: 0.5657\n",
            "Epoch 317/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7162 - acc: 0.7850 - val_loss: 1.4747 - val_acc: 0.6181\n",
            "Epoch 318/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7054 - acc: 0.7889 - val_loss: 1.4521 - val_acc: 0.6179\n",
            "Epoch 319/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 0.6909 - acc: 0.7920 - val_loss: 1.4565 - val_acc: 0.6181\n",
            "Epoch 320/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7142 - acc: 0.7827 - val_loss: 1.4764 - val_acc: 0.6100\n",
            "Epoch 321/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7173 - acc: 0.7889 - val_loss: 1.5220 - val_acc: 0.6138\n",
            "Epoch 322/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.6952 - acc: 0.7947 - val_loss: 1.4742 - val_acc: 0.6086\n",
            "Epoch 323/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7246 - acc: 0.7844 - val_loss: 1.4607 - val_acc: 0.6139\n",
            "Epoch 324/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.7032 - acc: 0.7953 - val_loss: 1.5494 - val_acc: 0.6039\n",
            "Epoch 325/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.6759 - acc: 0.7983 - val_loss: 1.4339 - val_acc: 0.6202\n",
            "Epoch 326/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 0.7016 - acc: 0.7900 - val_loss: 1.4900 - val_acc: 0.6154\n",
            "Epoch 327/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6598 - acc: 0.8041 - val_loss: 1.4454 - val_acc: 0.6245\n",
            "Epoch 328/350\n",
            "100/100 [==============================] - 12s 119ms/step - loss: 0.6680 - acc: 0.8009 - val_loss: 1.5204 - val_acc: 0.6051\n",
            "Epoch 329/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.6702 - acc: 0.8039 - val_loss: 1.4528 - val_acc: 0.6221\n",
            "Epoch 330/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.6727 - acc: 0.8030 - val_loss: 1.4942 - val_acc: 0.6089\n",
            "Epoch 331/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6936 - acc: 0.7898 - val_loss: 1.4506 - val_acc: 0.6189\n",
            "Epoch 332/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6797 - acc: 0.7972 - val_loss: 1.4395 - val_acc: 0.6251\n",
            "Epoch 333/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.6621 - acc: 0.8006 - val_loss: 1.4726 - val_acc: 0.6173\n",
            "Epoch 334/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6635 - acc: 0.8019 - val_loss: 1.5041 - val_acc: 0.6086\n",
            "Epoch 335/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.6844 - acc: 0.8031 - val_loss: 1.4794 - val_acc: 0.6205\n",
            "Epoch 336/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.6555 - acc: 0.8066 - val_loss: 1.4096 - val_acc: 0.6298\n",
            "Epoch 337/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6672 - acc: 0.8041 - val_loss: 1.4911 - val_acc: 0.6151\n",
            "Epoch 338/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6543 - acc: 0.8039 - val_loss: 1.4504 - val_acc: 0.6208\n",
            "Epoch 339/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.6670 - acc: 0.7992 - val_loss: 1.4653 - val_acc: 0.6246\n",
            "Epoch 340/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.6451 - acc: 0.8117 - val_loss: 1.4846 - val_acc: 0.6167\n",
            "Epoch 341/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6514 - acc: 0.8098 - val_loss: 1.5224 - val_acc: 0.6058\n",
            "Epoch 342/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6448 - acc: 0.8105 - val_loss: 1.4224 - val_acc: 0.6241\n",
            "Epoch 343/350\n",
            "100/100 [==============================] - 13s 128ms/step - loss: 0.6432 - acc: 0.8091 - val_loss: 1.4533 - val_acc: 0.6230\n",
            "Epoch 344/350\n",
            "100/100 [==============================] - 12s 118ms/step - loss: 0.6675 - acc: 0.8047 - val_loss: 1.5485 - val_acc: 0.6035\n",
            "Epoch 345/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6471 - acc: 0.8116 - val_loss: 1.4492 - val_acc: 0.6261\n",
            "Epoch 346/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6712 - acc: 0.7966 - val_loss: 1.4087 - val_acc: 0.6352\n",
            "Epoch 347/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6621 - acc: 0.8117 - val_loss: 1.4673 - val_acc: 0.6191\n",
            "Epoch 348/350\n",
            "100/100 [==============================] - 12s 117ms/step - loss: 0.6553 - acc: 0.8006 - val_loss: 1.4722 - val_acc: 0.6175\n",
            "Epoch 349/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.6566 - acc: 0.8055 - val_loss: 1.4967 - val_acc: 0.6119\n",
            "Epoch 350/350\n",
            "100/100 [==============================] - 12s 116ms/step - loss: 0.6458 - acc: 0.8086 - val_loss: 1.4374 - val_acc: 0.6235\n",
            "Training time: 5824.922385454178\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "8R-wq1uE-5Af",
        "outputId": "8a1f02a6-a307-495b-a47f-7323417c8502"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "URMcVG3A-7EL",
        "outputId": "dff801b0-3530-4a1f-8023-ad01bcbc3360"
      },
      "source": [
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 5s 14ms/step - loss: 1.4327 - acc: 0.6282\n",
            "accuracy on test set: acc of 62.81999945640564\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qZxc3whX2Ve5"
      },
      "source": [
        "##3.7 Batch norm model 4 - The best model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gl_Kvp9hpoND",
        "outputId": "ecdd0c65-7d80-4dcf-c9c3-85e974a4069f"
      },
      "source": [
        "# Batch norm model 4\n",
        "from keras.models import Sequential\n",
        "from keras.layers.convolutional import Conv2D\n",
        "from keras.layers.pooling import MaxPool2D\n",
        "from keras.layers.core import Dense,Activation,Dropout,Flatten\n",
        "\n",
        "from tensorflow.keras.layers import BatchNormalization\n",
        "from tensorflow.keras.initializers import RandomNormal, Constant\n",
        "model = Sequential()\n",
        " \n",
        "model.add(Conv2D(256,(3,3),padding='same',input_shape=(32,32,3)))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(256,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        " \n",
        "model.add(Conv2D(512,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(512,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "model.add(Conv2D(512,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(512,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "model.add(Conv2D(512,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Conv2D(512,(3,3),padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.2))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(1024))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(BatchNormalization(momentum=0.95, \n",
        "        epsilon=0.005,\n",
        "        beta_initializer=RandomNormal(mean=0.0, stddev=0.05), \n",
        "        gamma_initializer=Constant(value=0.9)))\n",
        "model.add(Dense(100,activation='softmax'))\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_4 (Conv2D)            (None, 32, 32, 256)       7168      \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 32, 32, 256)       1024      \n",
            "_________________________________________________________________\n",
            "activation (Activation)      (None, 32, 32, 256)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 32, 32, 256)       590080    \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 32, 32, 256)       1024      \n",
            "_________________________________________________________________\n",
            "activation_1 (Activation)    (None, 32, 32, 256)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 16, 16, 256)       0         \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 16, 16, 256)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_6 (Conv2D)            (None, 16, 16, 512)       1180160   \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 16, 16, 512)       2048      \n",
            "_________________________________________________________________\n",
            "activation_2 (Activation)    (None, 16, 16, 512)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 16, 16, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 16, 16, 512)       2048      \n",
            "_________________________________________________________________\n",
            "activation_3 (Activation)    (None, 16, 16, 512)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 8, 8, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 8, 8, 512)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 8, 8, 512)         2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_5 (Batch (None, 8, 8, 512)         2048      \n",
            "_________________________________________________________________\n",
            "activation_4 (Activation)    (None, 8, 8, 512)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 8, 8, 512)         2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_6 (Batch (None, 8, 8, 512)         2048      \n",
            "_________________________________________________________________\n",
            "activation_5 (Activation)    (None, 8, 8, 512)         0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_4 (MaxPooling2 (None, 4, 4, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 4, 4, 512)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_10 (Conv2D)           (None, 4, 4, 512)         2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_7 (Batch (None, 4, 4, 512)         2048      \n",
            "_________________________________________________________________\n",
            "activation_6 (Activation)    (None, 4, 4, 512)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_11 (Conv2D)           (None, 4, 4, 512)         2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_8 (Batch (None, 4, 4, 512)         2048      \n",
            "_________________________________________________________________\n",
            "activation_7 (Activation)    (None, 4, 4, 512)         0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_5 (MaxPooling2 (None, 2, 2, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_6 (Dropout)          (None, 2, 2, 512)         0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 2048)              0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 1024)              2098176   \n",
            "_________________________________________________________________\n",
            "activation_8 (Activation)    (None, 1024)              0         \n",
            "_________________________________________________________________\n",
            "dropout_7 (Dropout)          (None, 1024)              0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_9 (Batch (None, 1024)              4096      \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 100)               102500    \n",
            "=================================================================\n",
            "Total params: 15,795,556\n",
            "Trainable params: 15,786,340\n",
            "Non-trainable params: 9,216\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pOPc51VMp1kt"
      },
      "source": [
        "# Data Augmentation\n",
        "# Adding data augmentation for creating more images\n",
        "# Divide train and validation set \n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# Configuration for creating new images\n",
        "train_datagen = ImageDataGenerator(\n",
        "    rotation_range=20,\n",
        "    horizontal_flip=True,\n",
        ")\n",
        "\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(train_images, train_labels, test_size=0.2, random_state=93)\n",
        "train_datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kjuZKlzup7Xi"
      },
      "source": [
        "# Configure the model for training\n",
        "from tensorflow.keras import optimizers\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(learning_rate=1e-4),\n",
        "              metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2Jhkc8rI8RcK",
        "outputId": "32e32aff-4fe7-4801-c5d3-33fcf674b4a4"
      },
      "source": [
        "# Training model\n",
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_datagen.flow(X_train, y_train, batch_size=64),\n",
        "          steps_per_epoch=100, \n",
        "          epochs=350,\n",
        "          validation_data=(X_validation, y_validation),\n",
        "          verbose=1)\n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 40/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.8481 - acc: 0.4902 - val_loss: 2.0579 - val_acc: 0.4503\n",
            "Epoch 41/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.8030 - acc: 0.5105 - val_loss: 2.1553 - val_acc: 0.4360\n",
            "Epoch 42/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.7656 - acc: 0.5208 - val_loss: 2.0647 - val_acc: 0.4568\n",
            "Epoch 43/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.7834 - acc: 0.5148 - val_loss: 2.0246 - val_acc: 0.4623\n",
            "Epoch 44/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.7587 - acc: 0.5177 - val_loss: 2.0513 - val_acc: 0.4598\n",
            "Epoch 45/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.7126 - acc: 0.5259 - val_loss: 2.0400 - val_acc: 0.4552\n",
            "Epoch 46/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.7056 - acc: 0.5259 - val_loss: 2.1812 - val_acc: 0.4386\n",
            "Epoch 47/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.6794 - acc: 0.5391 - val_loss: 1.8853 - val_acc: 0.4925\n",
            "Epoch 48/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.6702 - acc: 0.5419 - val_loss: 1.7892 - val_acc: 0.5180\n",
            "Epoch 49/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.6663 - acc: 0.5417 - val_loss: 1.7869 - val_acc: 0.5157\n",
            "Epoch 50/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.6358 - acc: 0.5547 - val_loss: 2.3023 - val_acc: 0.4369\n",
            "Epoch 51/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.6325 - acc: 0.5495 - val_loss: 2.1002 - val_acc: 0.4535\n",
            "Epoch 52/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.6225 - acc: 0.5553 - val_loss: 1.8389 - val_acc: 0.5083\n",
            "Epoch 53/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.5973 - acc: 0.5536 - val_loss: 1.8487 - val_acc: 0.5070\n",
            "Epoch 54/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.5872 - acc: 0.5645 - val_loss: 1.9374 - val_acc: 0.4888\n",
            "Epoch 55/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.5555 - acc: 0.5661 - val_loss: 1.7048 - val_acc: 0.5320\n",
            "Epoch 56/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.5669 - acc: 0.5742 - val_loss: 1.7357 - val_acc: 0.5358\n",
            "Epoch 57/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.5226 - acc: 0.5766 - val_loss: 1.9109 - val_acc: 0.4970\n",
            "Epoch 58/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.5355 - acc: 0.5772 - val_loss: 1.9088 - val_acc: 0.4967\n",
            "Epoch 59/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.5305 - acc: 0.5741 - val_loss: 1.8395 - val_acc: 0.5131\n",
            "Epoch 60/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.4893 - acc: 0.5884 - val_loss: 1.8693 - val_acc: 0.5002\n",
            "Epoch 61/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.4884 - acc: 0.5931 - val_loss: 1.8360 - val_acc: 0.5115\n",
            "Epoch 62/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.4378 - acc: 0.6019 - val_loss: 1.7933 - val_acc: 0.5202\n",
            "Epoch 63/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.4134 - acc: 0.6078 - val_loss: 1.6906 - val_acc: 0.5342\n",
            "Epoch 64/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.4096 - acc: 0.6036 - val_loss: 1.7052 - val_acc: 0.5365\n",
            "Epoch 65/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.4252 - acc: 0.5983 - val_loss: 1.9660 - val_acc: 0.4876\n",
            "Epoch 66/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 1.4147 - acc: 0.6028 - val_loss: 1.7582 - val_acc: 0.5276\n",
            "Epoch 67/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.3926 - acc: 0.6072 - val_loss: 1.6726 - val_acc: 0.5498\n",
            "Epoch 68/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.3530 - acc: 0.6205 - val_loss: 1.8683 - val_acc: 0.5112\n",
            "Epoch 69/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.3903 - acc: 0.6130 - val_loss: 1.6327 - val_acc: 0.5534\n",
            "Epoch 70/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.3506 - acc: 0.6242 - val_loss: 1.6805 - val_acc: 0.5398\n",
            "Epoch 71/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.3482 - acc: 0.6227 - val_loss: 1.8856 - val_acc: 0.5069\n",
            "Epoch 72/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.3142 - acc: 0.6328 - val_loss: 1.8687 - val_acc: 0.5039\n",
            "Epoch 73/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.2837 - acc: 0.6427 - val_loss: 1.6982 - val_acc: 0.5464\n",
            "Epoch 74/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.2970 - acc: 0.6367 - val_loss: 1.9013 - val_acc: 0.5093\n",
            "Epoch 75/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.2818 - acc: 0.6408 - val_loss: 1.7457 - val_acc: 0.5311\n",
            "Epoch 76/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.2837 - acc: 0.6334 - val_loss: 1.6995 - val_acc: 0.5376\n",
            "Epoch 77/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.2677 - acc: 0.6442 - val_loss: 1.8436 - val_acc: 0.5127\n",
            "Epoch 78/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.2899 - acc: 0.6327 - val_loss: 1.6527 - val_acc: 0.5526\n",
            "Epoch 79/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.2720 - acc: 0.6461 - val_loss: 1.7320 - val_acc: 0.5402\n",
            "Epoch 80/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.2422 - acc: 0.6436 - val_loss: 1.7661 - val_acc: 0.5278\n",
            "Epoch 81/350\n",
            "100/100 [==============================] - 29s 290ms/step - loss: 1.2343 - acc: 0.6547 - val_loss: 1.7576 - val_acc: 0.5282\n",
            "Epoch 82/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.2203 - acc: 0.6472 - val_loss: 1.7886 - val_acc: 0.5284\n",
            "Epoch 83/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.2247 - acc: 0.6477 - val_loss: 1.5795 - val_acc: 0.5684\n",
            "Epoch 84/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.2302 - acc: 0.6536 - val_loss: 1.5887 - val_acc: 0.5648\n",
            "Epoch 85/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.1533 - acc: 0.6642 - val_loss: 1.6431 - val_acc: 0.5553\n",
            "Epoch 86/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.1781 - acc: 0.6628 - val_loss: 1.7791 - val_acc: 0.5231\n",
            "Epoch 87/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.1502 - acc: 0.6683 - val_loss: 1.7576 - val_acc: 0.5354\n",
            "Epoch 88/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.1566 - acc: 0.6775 - val_loss: 1.7090 - val_acc: 0.5422\n",
            "Epoch 89/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.1720 - acc: 0.6733 - val_loss: 1.6688 - val_acc: 0.5655\n",
            "Epoch 90/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.1429 - acc: 0.6747 - val_loss: 1.8088 - val_acc: 0.5327\n",
            "Epoch 91/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.1228 - acc: 0.6741 - val_loss: 1.6233 - val_acc: 0.5602\n",
            "Epoch 92/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.0948 - acc: 0.6870 - val_loss: 1.6122 - val_acc: 0.5693\n",
            "Epoch 93/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.1218 - acc: 0.6752 - val_loss: 1.5743 - val_acc: 0.5759\n",
            "Epoch 94/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.0886 - acc: 0.6917 - val_loss: 1.4617 - val_acc: 0.6031\n",
            "Epoch 95/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.0747 - acc: 0.6923 - val_loss: 1.7277 - val_acc: 0.5462\n",
            "Epoch 96/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.0833 - acc: 0.6869 - val_loss: 1.5690 - val_acc: 0.5753\n",
            "Epoch 97/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.1001 - acc: 0.6866 - val_loss: 1.5839 - val_acc: 0.5766\n",
            "Epoch 98/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.0632 - acc: 0.6950 - val_loss: 1.4883 - val_acc: 0.5930\n",
            "Epoch 99/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 1.0313 - acc: 0.7041 - val_loss: 1.5902 - val_acc: 0.5713\n",
            "Epoch 100/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.0738 - acc: 0.6889 - val_loss: 1.8055 - val_acc: 0.5298\n",
            "Epoch 101/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.0422 - acc: 0.6995 - val_loss: 1.6844 - val_acc: 0.5640\n",
            "Epoch 102/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.0313 - acc: 0.7028 - val_loss: 1.6949 - val_acc: 0.5605\n",
            "Epoch 103/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.0528 - acc: 0.6994 - val_loss: 1.4982 - val_acc: 0.5929\n",
            "Epoch 104/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9906 - acc: 0.7111 - val_loss: 1.5630 - val_acc: 0.5787\n",
            "Epoch 105/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.0007 - acc: 0.7141 - val_loss: 1.5780 - val_acc: 0.5833\n",
            "Epoch 106/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9913 - acc: 0.7111 - val_loss: 1.5451 - val_acc: 0.5888\n",
            "Epoch 107/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 1.0173 - acc: 0.7084 - val_loss: 1.5649 - val_acc: 0.5851\n",
            "Epoch 108/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9739 - acc: 0.7234 - val_loss: 1.5627 - val_acc: 0.5831\n",
            "Epoch 109/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 1.0050 - acc: 0.7023 - val_loss: 1.7768 - val_acc: 0.5460\n",
            "Epoch 110/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.9608 - acc: 0.7197 - val_loss: 1.6884 - val_acc: 0.5608\n",
            "Epoch 111/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9549 - acc: 0.7253 - val_loss: 1.4386 - val_acc: 0.6117\n",
            "Epoch 112/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.9337 - acc: 0.7367 - val_loss: 1.5674 - val_acc: 0.5866\n",
            "Epoch 113/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9476 - acc: 0.7228 - val_loss: 1.5347 - val_acc: 0.5914\n",
            "Epoch 114/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9539 - acc: 0.7223 - val_loss: 1.4251 - val_acc: 0.6176\n",
            "Epoch 115/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9593 - acc: 0.7186 - val_loss: 1.5034 - val_acc: 0.6012\n",
            "Epoch 116/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9162 - acc: 0.7389 - val_loss: 1.6104 - val_acc: 0.5752\n",
            "Epoch 117/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.9467 - acc: 0.7314 - val_loss: 1.6791 - val_acc: 0.5597\n",
            "Epoch 118/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9229 - acc: 0.7322 - val_loss: 1.5196 - val_acc: 0.5972\n",
            "Epoch 119/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.8813 - acc: 0.7464 - val_loss: 1.6294 - val_acc: 0.5743\n",
            "Epoch 120/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8710 - acc: 0.7458 - val_loss: 1.6248 - val_acc: 0.5788\n",
            "Epoch 121/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8886 - acc: 0.7411 - val_loss: 1.5289 - val_acc: 0.5970\n",
            "Epoch 122/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.9031 - acc: 0.7359 - val_loss: 1.5014 - val_acc: 0.6019\n",
            "Epoch 123/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8962 - acc: 0.7452 - val_loss: 1.5393 - val_acc: 0.5870\n",
            "Epoch 124/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8950 - acc: 0.7414 - val_loss: 1.4751 - val_acc: 0.6017\n",
            "Epoch 125/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.8492 - acc: 0.7570 - val_loss: 1.6252 - val_acc: 0.5776\n",
            "Epoch 126/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.8688 - acc: 0.7445 - val_loss: 1.6160 - val_acc: 0.5825\n",
            "Epoch 127/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.8510 - acc: 0.7530 - val_loss: 1.5109 - val_acc: 0.6020\n",
            "Epoch 128/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.8435 - acc: 0.7517 - val_loss: 1.4593 - val_acc: 0.6134\n",
            "Epoch 129/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8368 - acc: 0.7545 - val_loss: 1.5223 - val_acc: 0.5935\n",
            "Epoch 130/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8446 - acc: 0.7492 - val_loss: 1.5705 - val_acc: 0.5871\n",
            "Epoch 131/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.8105 - acc: 0.7581 - val_loss: 1.5301 - val_acc: 0.5948\n",
            "Epoch 132/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.8137 - acc: 0.7648 - val_loss: 1.5035 - val_acc: 0.6048\n",
            "Epoch 133/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7969 - acc: 0.7644 - val_loss: 1.5422 - val_acc: 0.5966\n",
            "Epoch 134/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.8060 - acc: 0.7591 - val_loss: 1.4680 - val_acc: 0.6091\n",
            "Epoch 135/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.7853 - acc: 0.7711 - val_loss: 1.5664 - val_acc: 0.5903\n",
            "Epoch 136/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7604 - acc: 0.7753 - val_loss: 1.5888 - val_acc: 0.5765\n",
            "Epoch 137/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7827 - acc: 0.7708 - val_loss: 1.4551 - val_acc: 0.6122\n",
            "Epoch 138/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7909 - acc: 0.7669 - val_loss: 1.5878 - val_acc: 0.5884\n",
            "Epoch 139/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7980 - acc: 0.7702 - val_loss: 1.4493 - val_acc: 0.6173\n",
            "Epoch 140/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.7988 - acc: 0.7609 - val_loss: 1.5778 - val_acc: 0.5857\n",
            "Epoch 141/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7556 - acc: 0.7812 - val_loss: 1.4958 - val_acc: 0.6026\n",
            "Epoch 142/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7657 - acc: 0.7777 - val_loss: 1.6430 - val_acc: 0.5862\n",
            "Epoch 143/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.7497 - acc: 0.7830 - val_loss: 1.7227 - val_acc: 0.5607\n",
            "Epoch 144/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7573 - acc: 0.7706 - val_loss: 1.5078 - val_acc: 0.6131\n",
            "Epoch 145/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.7421 - acc: 0.7816 - val_loss: 1.4629 - val_acc: 0.6137\n",
            "Epoch 146/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.7513 - acc: 0.7794 - val_loss: 1.5593 - val_acc: 0.5978\n",
            "Epoch 147/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7328 - acc: 0.7855 - val_loss: 1.5556 - val_acc: 0.5992\n",
            "Epoch 148/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.7204 - acc: 0.7873 - val_loss: 1.4432 - val_acc: 0.6191\n",
            "Epoch 149/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.7308 - acc: 0.7820 - val_loss: 1.5641 - val_acc: 0.5952\n",
            "Epoch 150/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.7365 - acc: 0.7822 - val_loss: 1.4579 - val_acc: 0.6163\n",
            "Epoch 151/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.6968 - acc: 0.7953 - val_loss: 1.4835 - val_acc: 0.6153\n",
            "Epoch 152/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.7035 - acc: 0.7966 - val_loss: 1.4915 - val_acc: 0.6035\n",
            "Epoch 153/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.6962 - acc: 0.7911 - val_loss: 1.4898 - val_acc: 0.6116\n",
            "Epoch 154/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.7152 - acc: 0.7914 - val_loss: 1.6388 - val_acc: 0.5892\n",
            "Epoch 155/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.7109 - acc: 0.7875 - val_loss: 1.5215 - val_acc: 0.6133\n",
            "Epoch 156/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6829 - acc: 0.7994 - val_loss: 1.5013 - val_acc: 0.6146\n",
            "Epoch 157/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6937 - acc: 0.7978 - val_loss: 1.4649 - val_acc: 0.6195\n",
            "Epoch 158/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6761 - acc: 0.8002 - val_loss: 1.5033 - val_acc: 0.6123\n",
            "Epoch 159/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6757 - acc: 0.8058 - val_loss: 1.4555 - val_acc: 0.6218\n",
            "Epoch 160/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6638 - acc: 0.8067 - val_loss: 1.7982 - val_acc: 0.5596\n",
            "Epoch 161/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6876 - acc: 0.7958 - val_loss: 1.5536 - val_acc: 0.5943\n",
            "Epoch 162/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.6500 - acc: 0.8073 - val_loss: 1.5337 - val_acc: 0.6089\n",
            "Epoch 163/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6482 - acc: 0.8086 - val_loss: 1.7489 - val_acc: 0.5760\n",
            "Epoch 164/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.6747 - acc: 0.7983 - val_loss: 1.3666 - val_acc: 0.6437\n",
            "Epoch 165/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.6287 - acc: 0.8131 - val_loss: 1.5351 - val_acc: 0.6087\n",
            "Epoch 166/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.6243 - acc: 0.8130 - val_loss: 1.5239 - val_acc: 0.6065\n",
            "Epoch 167/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6207 - acc: 0.8131 - val_loss: 1.4618 - val_acc: 0.6233\n",
            "Epoch 168/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.6097 - acc: 0.8169 - val_loss: 1.4852 - val_acc: 0.6254\n",
            "Epoch 169/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6209 - acc: 0.8161 - val_loss: 1.5286 - val_acc: 0.6084\n",
            "Epoch 170/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.6338 - acc: 0.8114 - val_loss: 1.4589 - val_acc: 0.6262\n",
            "Epoch 171/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5970 - acc: 0.8245 - val_loss: 1.4512 - val_acc: 0.6260\n",
            "Epoch 172/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.6223 - acc: 0.8169 - val_loss: 1.4354 - val_acc: 0.6335\n",
            "Epoch 173/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6099 - acc: 0.8164 - val_loss: 1.5272 - val_acc: 0.6126\n",
            "Epoch 174/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6082 - acc: 0.8167 - val_loss: 1.4740 - val_acc: 0.6220\n",
            "Epoch 175/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5972 - acc: 0.8230 - val_loss: 1.4423 - val_acc: 0.6347\n",
            "Epoch 176/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.6006 - acc: 0.8197 - val_loss: 1.4709 - val_acc: 0.6302\n",
            "Epoch 177/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5765 - acc: 0.8273 - val_loss: 1.6027 - val_acc: 0.6036\n",
            "Epoch 178/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5937 - acc: 0.8202 - val_loss: 1.4956 - val_acc: 0.6192\n",
            "Epoch 179/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.6040 - acc: 0.8181 - val_loss: 1.4778 - val_acc: 0.6250\n",
            "Epoch 180/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5662 - acc: 0.8292 - val_loss: 1.4592 - val_acc: 0.6312\n",
            "Epoch 181/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5861 - acc: 0.8213 - val_loss: 1.4868 - val_acc: 0.6253\n",
            "Epoch 182/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5555 - acc: 0.8314 - val_loss: 1.6554 - val_acc: 0.5960\n",
            "Epoch 183/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5370 - acc: 0.8367 - val_loss: 1.5546 - val_acc: 0.6144\n",
            "Epoch 184/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5488 - acc: 0.8384 - val_loss: 1.4965 - val_acc: 0.6246\n",
            "Epoch 185/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.5651 - acc: 0.8313 - val_loss: 1.4495 - val_acc: 0.6291\n",
            "Epoch 186/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5557 - acc: 0.8333 - val_loss: 1.5143 - val_acc: 0.6206\n",
            "Epoch 187/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5332 - acc: 0.8370 - val_loss: 1.6701 - val_acc: 0.5995\n",
            "Epoch 188/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5279 - acc: 0.8391 - val_loss: 1.5621 - val_acc: 0.6049\n",
            "Epoch 189/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5377 - acc: 0.8338 - val_loss: 1.5542 - val_acc: 0.6133\n",
            "Epoch 190/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.5575 - acc: 0.8328 - val_loss: 1.4372 - val_acc: 0.6328\n",
            "Epoch 191/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5201 - acc: 0.8434 - val_loss: 1.5671 - val_acc: 0.6116\n",
            "Epoch 192/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5328 - acc: 0.8408 - val_loss: 1.4976 - val_acc: 0.6159\n",
            "Epoch 193/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5109 - acc: 0.8453 - val_loss: 1.5577 - val_acc: 0.6077\n",
            "Epoch 194/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.5286 - acc: 0.8452 - val_loss: 1.4678 - val_acc: 0.6295\n",
            "Epoch 195/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5179 - acc: 0.8452 - val_loss: 1.4924 - val_acc: 0.6302\n",
            "Epoch 196/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5234 - acc: 0.8417 - val_loss: 1.5445 - val_acc: 0.6183\n",
            "Epoch 197/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.5105 - acc: 0.8434 - val_loss: 1.5094 - val_acc: 0.6210\n",
            "Epoch 198/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4928 - acc: 0.8572 - val_loss: 1.4775 - val_acc: 0.6356\n",
            "Epoch 199/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5028 - acc: 0.8478 - val_loss: 1.4718 - val_acc: 0.6203\n",
            "Epoch 200/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4744 - acc: 0.8580 - val_loss: 1.5726 - val_acc: 0.6117\n",
            "Epoch 201/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4959 - acc: 0.8478 - val_loss: 1.6163 - val_acc: 0.6097\n",
            "Epoch 202/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.5105 - acc: 0.8450 - val_loss: 1.5231 - val_acc: 0.6219\n",
            "Epoch 203/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4802 - acc: 0.8544 - val_loss: 1.4153 - val_acc: 0.6507\n",
            "Epoch 204/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4876 - acc: 0.8511 - val_loss: 1.6175 - val_acc: 0.6064\n",
            "Epoch 205/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4962 - acc: 0.8489 - val_loss: 1.4433 - val_acc: 0.6442\n",
            "Epoch 206/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4783 - acc: 0.8545 - val_loss: 1.5717 - val_acc: 0.6187\n",
            "Epoch 207/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.4641 - acc: 0.8595 - val_loss: 1.4895 - val_acc: 0.6381\n",
            "Epoch 208/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.4704 - acc: 0.8587 - val_loss: 1.5478 - val_acc: 0.6253\n",
            "Epoch 209/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4738 - acc: 0.8550 - val_loss: 1.4709 - val_acc: 0.6327\n",
            "Epoch 210/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4728 - acc: 0.8545 - val_loss: 1.5849 - val_acc: 0.6098\n",
            "Epoch 211/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4591 - acc: 0.8587 - val_loss: 1.5073 - val_acc: 0.6231\n",
            "Epoch 212/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.4486 - acc: 0.8633 - val_loss: 1.5223 - val_acc: 0.6291\n",
            "Epoch 213/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4598 - acc: 0.8600 - val_loss: 1.4897 - val_acc: 0.6354\n",
            "Epoch 214/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4491 - acc: 0.8630 - val_loss: 1.4919 - val_acc: 0.6360\n",
            "Epoch 215/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4517 - acc: 0.8605 - val_loss: 1.6241 - val_acc: 0.6118\n",
            "Epoch 216/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4524 - acc: 0.8594 - val_loss: 1.4949 - val_acc: 0.6319\n",
            "Epoch 217/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4492 - acc: 0.8656 - val_loss: 1.5055 - val_acc: 0.6311\n",
            "Epoch 218/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4340 - acc: 0.8708 - val_loss: 1.5506 - val_acc: 0.6276\n",
            "Epoch 219/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.4157 - acc: 0.8737 - val_loss: 1.5824 - val_acc: 0.6219\n",
            "Epoch 220/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.4530 - acc: 0.8612 - val_loss: 1.4988 - val_acc: 0.6315\n",
            "Epoch 221/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.4168 - acc: 0.8733 - val_loss: 1.4733 - val_acc: 0.6425\n",
            "Epoch 222/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4437 - acc: 0.8627 - val_loss: 1.5193 - val_acc: 0.6220\n",
            "Epoch 223/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4287 - acc: 0.8675 - val_loss: 1.6140 - val_acc: 0.6075\n",
            "Epoch 224/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4370 - acc: 0.8694 - val_loss: 1.5824 - val_acc: 0.6190\n",
            "Epoch 225/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.4210 - acc: 0.8764 - val_loss: 1.5243 - val_acc: 0.6310\n",
            "Epoch 226/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4103 - acc: 0.8722 - val_loss: 1.4686 - val_acc: 0.6438\n",
            "Epoch 227/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.4097 - acc: 0.8791 - val_loss: 1.5215 - val_acc: 0.6328\n",
            "Epoch 228/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4165 - acc: 0.8714 - val_loss: 1.5185 - val_acc: 0.6366\n",
            "Epoch 229/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3896 - acc: 0.8839 - val_loss: 1.6233 - val_acc: 0.6205\n",
            "Epoch 230/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3960 - acc: 0.8783 - val_loss: 1.4840 - val_acc: 0.6420\n",
            "Epoch 231/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3927 - acc: 0.8809 - val_loss: 1.5066 - val_acc: 0.6332\n",
            "Epoch 232/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3871 - acc: 0.8792 - val_loss: 1.4460 - val_acc: 0.6483\n",
            "Epoch 233/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.4105 - acc: 0.8766 - val_loss: 1.6578 - val_acc: 0.6070\n",
            "Epoch 234/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3747 - acc: 0.8856 - val_loss: 1.5027 - val_acc: 0.6409\n",
            "Epoch 235/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3925 - acc: 0.8775 - val_loss: 1.5166 - val_acc: 0.6394\n",
            "Epoch 236/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3865 - acc: 0.8803 - val_loss: 1.4999 - val_acc: 0.6452\n",
            "Epoch 237/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3892 - acc: 0.8794 - val_loss: 1.5460 - val_acc: 0.6337\n",
            "Epoch 238/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3771 - acc: 0.8852 - val_loss: 1.5627 - val_acc: 0.6288\n",
            "Epoch 239/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3824 - acc: 0.8797 - val_loss: 1.5586 - val_acc: 0.6231\n",
            "Epoch 240/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.3638 - acc: 0.8895 - val_loss: 1.5398 - val_acc: 0.6321\n",
            "Epoch 241/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3912 - acc: 0.8791 - val_loss: 1.5968 - val_acc: 0.6181\n",
            "Epoch 242/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3857 - acc: 0.8858 - val_loss: 1.5360 - val_acc: 0.6274\n",
            "Epoch 243/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3803 - acc: 0.8838 - val_loss: 1.5249 - val_acc: 0.6369\n",
            "Epoch 244/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3670 - acc: 0.8842 - val_loss: 1.5765 - val_acc: 0.6278\n",
            "Epoch 245/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3860 - acc: 0.8845 - val_loss: 1.5295 - val_acc: 0.6395\n",
            "Epoch 246/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3628 - acc: 0.8903 - val_loss: 1.5429 - val_acc: 0.6327\n",
            "Epoch 247/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3681 - acc: 0.8923 - val_loss: 1.4756 - val_acc: 0.6466\n",
            "Epoch 248/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3668 - acc: 0.8866 - val_loss: 1.5698 - val_acc: 0.6288\n",
            "Epoch 249/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3634 - acc: 0.8894 - val_loss: 1.5567 - val_acc: 0.6358\n",
            "Epoch 250/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3504 - acc: 0.8905 - val_loss: 1.6380 - val_acc: 0.6223\n",
            "Epoch 251/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.3605 - acc: 0.8928 - val_loss: 1.6064 - val_acc: 0.6219\n",
            "Epoch 252/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3624 - acc: 0.8914 - val_loss: 1.6023 - val_acc: 0.6273\n",
            "Epoch 253/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3557 - acc: 0.8888 - val_loss: 1.5444 - val_acc: 0.6299\n",
            "Epoch 254/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3562 - acc: 0.8914 - val_loss: 1.4813 - val_acc: 0.6483\n",
            "Epoch 255/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3658 - acc: 0.8895 - val_loss: 1.5492 - val_acc: 0.6425\n",
            "Epoch 256/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3336 - acc: 0.8986 - val_loss: 1.4721 - val_acc: 0.6495\n",
            "Epoch 257/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3527 - acc: 0.8903 - val_loss: 1.5728 - val_acc: 0.6368\n",
            "Epoch 258/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.3514 - acc: 0.8914 - val_loss: 1.6017 - val_acc: 0.6256\n",
            "Epoch 259/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3661 - acc: 0.8900 - val_loss: 1.5369 - val_acc: 0.6401\n",
            "Epoch 260/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3272 - acc: 0.8986 - val_loss: 1.6783 - val_acc: 0.6119\n",
            "Epoch 261/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.3383 - acc: 0.8959 - val_loss: 1.5007 - val_acc: 0.6479\n",
            "Epoch 262/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.3308 - acc: 0.8953 - val_loss: 1.6498 - val_acc: 0.6186\n",
            "Epoch 263/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.3315 - acc: 0.8978 - val_loss: 1.4891 - val_acc: 0.6546\n",
            "Epoch 264/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3251 - acc: 0.9028 - val_loss: 1.5636 - val_acc: 0.6340\n",
            "Epoch 265/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3251 - acc: 0.8980 - val_loss: 1.5714 - val_acc: 0.6314\n",
            "Epoch 266/350\n",
            "100/100 [==============================] - 29s 291ms/step - loss: 0.3259 - acc: 0.9000 - val_loss: 1.5979 - val_acc: 0.6232\n",
            "Epoch 267/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3357 - acc: 0.8995 - val_loss: 1.4896 - val_acc: 0.6476\n",
            "Epoch 268/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3304 - acc: 0.8986 - val_loss: 1.6236 - val_acc: 0.6253\n",
            "Epoch 269/350\n",
            "100/100 [==============================] - 29s 292ms/step - loss: 0.3171 - acc: 0.9017 - val_loss: 1.5661 - val_acc: 0.6319\n",
            "Epoch 270/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3169 - acc: 0.9005 - val_loss: 1.5163 - val_acc: 0.6467\n",
            "Epoch 271/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3179 - acc: 0.9078 - val_loss: 1.6190 - val_acc: 0.6287\n",
            "Epoch 272/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.3064 - acc: 0.9086 - val_loss: 1.5402 - val_acc: 0.6412\n",
            "Epoch 273/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.3060 - acc: 0.9062 - val_loss: 1.5972 - val_acc: 0.6337\n",
            "Epoch 274/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3139 - acc: 0.9019 - val_loss: 1.5314 - val_acc: 0.6449\n",
            "Epoch 275/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3016 - acc: 0.9053 - val_loss: 1.4934 - val_acc: 0.6494\n",
            "Epoch 276/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3266 - acc: 0.8977 - val_loss: 1.5957 - val_acc: 0.6320\n",
            "Epoch 277/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.3085 - acc: 0.9034 - val_loss: 1.5055 - val_acc: 0.6505\n",
            "Epoch 278/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.3045 - acc: 0.9042 - val_loss: 1.5466 - val_acc: 0.6399\n",
            "Epoch 279/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.2942 - acc: 0.9137 - val_loss: 1.6064 - val_acc: 0.6353\n",
            "Epoch 280/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2951 - acc: 0.9095 - val_loss: 1.6020 - val_acc: 0.6338\n",
            "Epoch 281/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2960 - acc: 0.9106 - val_loss: 1.5592 - val_acc: 0.6487\n",
            "Epoch 282/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2964 - acc: 0.9084 - val_loss: 1.6844 - val_acc: 0.6180\n",
            "Epoch 283/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2925 - acc: 0.9136 - val_loss: 1.6026 - val_acc: 0.6353\n",
            "Epoch 284/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2899 - acc: 0.9117 - val_loss: 1.6017 - val_acc: 0.6355\n",
            "Epoch 285/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.3018 - acc: 0.9031 - val_loss: 1.5779 - val_acc: 0.6396\n",
            "Epoch 286/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2696 - acc: 0.9175 - val_loss: 1.6101 - val_acc: 0.6274\n",
            "Epoch 287/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2787 - acc: 0.9145 - val_loss: 1.5164 - val_acc: 0.6492\n",
            "Epoch 288/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2806 - acc: 0.9139 - val_loss: 1.5947 - val_acc: 0.6349\n",
            "Epoch 289/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2836 - acc: 0.9120 - val_loss: 1.5694 - val_acc: 0.6398\n",
            "Epoch 290/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2873 - acc: 0.9111 - val_loss: 1.5838 - val_acc: 0.6393\n",
            "Epoch 291/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2827 - acc: 0.9147 - val_loss: 1.6727 - val_acc: 0.6219\n",
            "Epoch 292/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2692 - acc: 0.9169 - val_loss: 1.6244 - val_acc: 0.6305\n",
            "Epoch 293/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2777 - acc: 0.9156 - val_loss: 1.6099 - val_acc: 0.6317\n",
            "Epoch 294/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2786 - acc: 0.9180 - val_loss: 1.5958 - val_acc: 0.6355\n",
            "Epoch 295/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2895 - acc: 0.9125 - val_loss: 1.5848 - val_acc: 0.6431\n",
            "Epoch 296/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2784 - acc: 0.9127 - val_loss: 1.5837 - val_acc: 0.6418\n",
            "Epoch 297/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2736 - acc: 0.9148 - val_loss: 1.5572 - val_acc: 0.6405\n",
            "Epoch 298/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2755 - acc: 0.9178 - val_loss: 1.7080 - val_acc: 0.6210\n",
            "Epoch 299/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2699 - acc: 0.9177 - val_loss: 1.6061 - val_acc: 0.6372\n",
            "Epoch 300/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2740 - acc: 0.9183 - val_loss: 1.5602 - val_acc: 0.6424\n",
            "Epoch 301/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2730 - acc: 0.9183 - val_loss: 1.6649 - val_acc: 0.6373\n",
            "Epoch 302/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2738 - acc: 0.9195 - val_loss: 1.5581 - val_acc: 0.6433\n",
            "Epoch 303/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2759 - acc: 0.9177 - val_loss: 1.5809 - val_acc: 0.6478\n",
            "Epoch 304/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2486 - acc: 0.9245 - val_loss: 1.6166 - val_acc: 0.6327\n",
            "Epoch 305/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2549 - acc: 0.9237 - val_loss: 1.5569 - val_acc: 0.6468\n",
            "Epoch 306/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2602 - acc: 0.9172 - val_loss: 1.6380 - val_acc: 0.6379\n",
            "Epoch 307/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2514 - acc: 0.9214 - val_loss: 1.5527 - val_acc: 0.6503\n",
            "Epoch 308/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2535 - acc: 0.9198 - val_loss: 1.5517 - val_acc: 0.6498\n",
            "Epoch 309/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2553 - acc: 0.9202 - val_loss: 1.5093 - val_acc: 0.6510\n",
            "Epoch 310/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.2609 - acc: 0.9248 - val_loss: 1.6440 - val_acc: 0.6272\n",
            "Epoch 311/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2707 - acc: 0.9203 - val_loss: 1.5653 - val_acc: 0.6506\n",
            "Epoch 312/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2526 - acc: 0.9216 - val_loss: 1.6174 - val_acc: 0.6384\n",
            "Epoch 313/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2489 - acc: 0.9275 - val_loss: 1.6292 - val_acc: 0.6397\n",
            "Epoch 314/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2452 - acc: 0.9273 - val_loss: 1.5483 - val_acc: 0.6493\n",
            "Epoch 315/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2615 - acc: 0.9191 - val_loss: 1.6338 - val_acc: 0.6453\n",
            "Epoch 316/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2590 - acc: 0.9216 - val_loss: 1.6336 - val_acc: 0.6304\n",
            "Epoch 317/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2501 - acc: 0.9194 - val_loss: 1.5871 - val_acc: 0.6496\n",
            "Epoch 318/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2412 - acc: 0.9291 - val_loss: 1.5151 - val_acc: 0.6525\n",
            "Epoch 319/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2426 - acc: 0.9264 - val_loss: 1.5733 - val_acc: 0.6461\n",
            "Epoch 320/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2316 - acc: 0.9300 - val_loss: 1.6128 - val_acc: 0.6455\n",
            "Epoch 321/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2516 - acc: 0.9223 - val_loss: 1.5718 - val_acc: 0.6506\n",
            "Epoch 322/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2328 - acc: 0.9286 - val_loss: 1.6932 - val_acc: 0.6270\n",
            "Epoch 323/350\n",
            "100/100 [==============================] - 30s 295ms/step - loss: 0.2336 - acc: 0.9295 - val_loss: 1.5650 - val_acc: 0.6464\n",
            "Epoch 324/350\n",
            "100/100 [==============================] - 29s 296ms/step - loss: 0.2405 - acc: 0.9237 - val_loss: 1.5952 - val_acc: 0.6387\n",
            "Epoch 325/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2270 - acc: 0.9298 - val_loss: 1.5697 - val_acc: 0.6454\n",
            "Epoch 326/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2536 - acc: 0.9169 - val_loss: 1.5860 - val_acc: 0.6467\n",
            "Epoch 327/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2404 - acc: 0.9253 - val_loss: 1.5752 - val_acc: 0.6453\n",
            "Epoch 328/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2427 - acc: 0.9242 - val_loss: 1.7094 - val_acc: 0.6279\n",
            "Epoch 329/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.2280 - acc: 0.9298 - val_loss: 1.6128 - val_acc: 0.6489\n",
            "Epoch 330/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2292 - acc: 0.9302 - val_loss: 1.6261 - val_acc: 0.6441\n",
            "Epoch 331/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2375 - acc: 0.9292 - val_loss: 1.5269 - val_acc: 0.6560\n",
            "Epoch 332/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2254 - acc: 0.9308 - val_loss: 1.5537 - val_acc: 0.6526\n",
            "Epoch 333/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2272 - acc: 0.9306 - val_loss: 1.5623 - val_acc: 0.6547\n",
            "Epoch 334/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2288 - acc: 0.9308 - val_loss: 1.5828 - val_acc: 0.6501\n",
            "Epoch 335/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2424 - acc: 0.9247 - val_loss: 1.5259 - val_acc: 0.6569\n",
            "Epoch 336/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2175 - acc: 0.9348 - val_loss: 1.5797 - val_acc: 0.6554\n",
            "Epoch 337/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2073 - acc: 0.9336 - val_loss: 1.7479 - val_acc: 0.6203\n",
            "Epoch 338/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2250 - acc: 0.9286 - val_loss: 1.5088 - val_acc: 0.6557\n",
            "Epoch 339/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2113 - acc: 0.9358 - val_loss: 1.5235 - val_acc: 0.6574\n",
            "Epoch 340/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2248 - acc: 0.9312 - val_loss: 1.6319 - val_acc: 0.6365\n",
            "Epoch 341/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2209 - acc: 0.9322 - val_loss: 1.5662 - val_acc: 0.6484\n",
            "Epoch 342/350\n",
            "100/100 [==============================] - 29s 293ms/step - loss: 0.2195 - acc: 0.9331 - val_loss: 1.5373 - val_acc: 0.6571\n",
            "Epoch 343/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.1984 - acc: 0.9403 - val_loss: 1.6281 - val_acc: 0.6481\n",
            "Epoch 344/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2211 - acc: 0.9347 - val_loss: 1.6306 - val_acc: 0.6507\n",
            "Epoch 345/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2060 - acc: 0.9352 - val_loss: 1.5721 - val_acc: 0.6537\n",
            "Epoch 346/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2190 - acc: 0.9347 - val_loss: 1.6891 - val_acc: 0.6388\n",
            "Epoch 347/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2255 - acc: 0.9281 - val_loss: 1.6063 - val_acc: 0.6525\n",
            "Epoch 348/350\n",
            "100/100 [==============================] - 29s 294ms/step - loss: 0.2197 - acc: 0.9308 - val_loss: 1.6667 - val_acc: 0.6393\n",
            "Epoch 349/350\n",
            "100/100 [==============================] - 29s 295ms/step - loss: 0.2201 - acc: 0.9303 - val_loss: 1.6235 - val_acc: 0.6453\n",
            "Epoch 350/350\n",
            "100/100 [==============================] - 30s 296ms/step - loss: 0.2009 - acc: 0.9405 - val_loss: 1.5574 - val_acc: 0.6584\n",
            "Training time: 12537.93370604515\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "Pb5KWmzY8UTT",
        "outputId": "5edaf24e-9b16-4d57-cb24-3cf7551df879"
      },
      "source": [
        "# Visualize training process\n",
        "training_plot(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QBwz7T868V59",
        "outputId": "f123185c-59e2-4195-de48-a7b7f640edd4"
      },
      "source": [
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 11s 35ms/step - loss: 1.5613 - acc: 0.6615\n",
            "accuracy on test set: acc of 66.14999771118164\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0iBdZ8_l3uGc",
        "outputId": "f574cbbf-3dfc-478d-d098-0656dd219b69"
      },
      "source": [
        "from google.colab import drive\n",
        "\n",
        "ROOT = \"/content/drive\"\n",
        "print(ROOT)\n",
        "drive.mount(ROOT)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/drive\n",
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "whVkI-kCblSr",
        "outputId": "8fae0ef9-a07d-4bdb-b72c-51125cce7cd6"
      },
      "source": [
        "%cd drive/MyDrive/results/"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/drive/MyDrive/results\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qr7ilLze357K"
      },
      "source": [
        "!pwd\n",
        "model.save('66%_accuracy_on_test.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ihGdsmGR4MzP"
      },
      "source": [
        "from keras.models import load_model\n",
        "model = load_model('66%_accuracy_on_test.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LK7cGc1Z4S5W",
        "outputId": "8ff81dfc-fcc1-4f38-d371-356e15f2b129"
      },
      "source": [
        "scores = model.evaluate(test_images, test_labels)\n",
        "print(f'accuracy on test set: {model.metrics_names[1]} of {scores[1]*100}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 41s 37ms/step - loss: 1.5613 - acc: 0.6615\n",
            "accuracy on test set: acc of 66.14999771118164\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0sjAMgPWcHFJ",
        "outputId": "ea2b9763-6b85-49a5-caff-343d78ab1c0a"
      },
      "source": [
        "# Translate categorial to array for drawing confusion matrix\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from numpy import argmax\n",
        "prediction = []\n",
        "true_labels = []\n",
        "\n",
        "pred = model.predict(test_images)\n",
        "print(test_labels.shape[0])\n",
        "for i in range(test_labels.shape[0]):\n",
        "  prediction.append(argmax(pred[i]))\n",
        "  true_labels.append(argmax(test_labels[i]))\n",
        "\n",
        "cm = confusion_matrix(prediction, true_labels)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "10000\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1MHRpKW_lv-_"
      },
      "source": [
        "# Name of all classes in CIFAR-100\n",
        "classes = ['beaver', 'dolphin', 'otter', 'seal', 'whale', \n",
        "'aquarium' ,'fish', 'ray', 'shark', 'trout', \n",
        "'orchids', 'poppies', 'roses', 'sunflowers', 'tulips', \n",
        "'bottles', 'bowls', 'cans', 'cups', 'plates', \n",
        "'apples', 'mushrooms', 'oranges', 'pears', 'sweet peppers', \n",
        "'clock', 'computer keyboard', 'lamp', 'telephone', 'television', 'bed', 'chair', 'couch', 'table', 'wardrobe', \n",
        "'bee', 'beetle', 'butterfly', 'caterpillar', 'cockroach', \n",
        "'bear', 'leopard', 'lion', 'tiger', 'wolf', \n",
        "'bridge', 'castle', 'house', 'road', 'skyscraper', \n",
        "'cloud', 'forest', 'mountain', 'plain', 'sea', \n",
        "'camel', 'cattle', 'chimpanzee', 'elephant', 'kangaroo', \n",
        "'fox', 'porcupine', 'possum', 'raccoon', 'skunk', \n",
        "'crab', 'lobster', 'snail', 'spider', 'worm', \n",
        "'baby', 'boy', 'girl', 'man', 'woman', \n",
        "'crocodile', 'dinosaur', 'lizard', 'snake', 'turtle', \n",
        "'hamster', 'mouse', 'rabbit', 'shrew', 'squirrel', \n",
        "'maple', 'oak', 'palm', 'pine', 'willow', \n",
        "'bicycle', 'bus', 'motorcycle', 'pickup truck', 'train', \n",
        "'lawn-mower', 'rocket', 'streetcar', 'tank', 'tractor']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 803
        },
        "id": "yf2H0enCio_d",
        "outputId": "380918f8-de54-40c3-8457-67b429d34235"
      },
      "source": [
        "# Plot the confusion matrix\n",
        "import matplotlib.pyplot as plt\n",
        "print(cm)\n",
        "fig = plt.figure(figsize=(24,24))\n",
        "ax = fig.add_subplot(211)\n",
        "cax = ax.matshow(cm)\n",
        "plt.title('Confusion matrix of the classifier')\n",
        "fig.colorbar(cax)\n",
        "ax.set_xticklabels([''] + classes)\n",
        "ax.set_yticklabels([''] + classes)\n",
        "plt.xlabel('Predicted')\n",
        "plt.ylabel('True')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[67  0  0 ...  0  0  0]\n",
            " [ 0 85  1 ...  0  1  0]\n",
            " [ 0  0 38 ...  0  4  0]\n",
            " ...\n",
            " [ 0  0  0 ... 59  0  0]\n",
            " [ 0  0  4 ...  0 41  0]\n",
            " [ 0  0  0 ...  0  0 67]]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1728x1728 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5hoLUnNF4OzB",
        "outputId": "c772f796-2b87-46bf-e843-7ed42c0eb106"
      },
      "source": [
        "# Calculating f1 score\n",
        "from sklearn.metrics import f1_score\n",
        "print(f\"f1 score: {f1_score(true_labels, prediction, average='weighted')}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "f1 score: 0.6636194116852772\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4azi3XF95Y71",
        "outputId": "1d72c908-1e48-4808-8f57-9a3caf396f8e"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "print(classification_report(true_labels, prediction, target_names=classes, digits=5))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "                   precision    recall  f1-score   support\n",
            "\n",
            "           beaver    0.93056   0.67000   0.77907       100\n",
            "          dolphin    0.74561   0.85000   0.79439       100\n",
            "            otter    0.70370   0.38000   0.49351       100\n",
            "             seal    0.47899   0.57000   0.52055       100\n",
            "            whale    0.35912   0.65000   0.46263       100\n",
            "         aquarium    0.63025   0.75000   0.68493       100\n",
            "             fish    0.66142   0.84000   0.74009       100\n",
            "              ray    0.80000   0.72000   0.75789       100\n",
            "            shark    0.97101   0.67000   0.79290       100\n",
            "            trout    0.81609   0.71000   0.75936       100\n",
            "          orchids    0.60714   0.51000   0.55435       100\n",
            "          poppies    0.55294   0.47000   0.50811       100\n",
            "            roses    0.63359   0.83000   0.71861       100\n",
            "       sunflowers    0.58824   0.70000   0.63927       100\n",
            "           tulips    0.59259   0.64000   0.61538       100\n",
            "          bottles    0.66316   0.63000   0.64615       100\n",
            "            bowls    0.74038   0.77000   0.75490       100\n",
            "             cans    0.85057   0.74000   0.79144       100\n",
            "             cups    0.63043   0.58000   0.60417       100\n",
            "           plates    0.70000   0.63000   0.66316       100\n",
            "           apples    0.88660   0.86000   0.87310       100\n",
            "        mushrooms    0.95652   0.66000   0.78107       100\n",
            "          oranges    0.78205   0.61000   0.68539       100\n",
            "            pears    0.90667   0.68000   0.77714       100\n",
            "    sweet peppers    0.94444   0.68000   0.79070       100\n",
            "            clock    0.65672   0.44000   0.52695       100\n",
            "computer keyboard    0.78205   0.61000   0.68539       100\n",
            "             lamp    0.50000   0.56000   0.52830       100\n",
            "        telephone    0.80412   0.78000   0.79188       100\n",
            "       television    0.72527   0.66000   0.69110       100\n",
            "              bed    0.72222   0.52000   0.60465       100\n",
            "            chair    0.81690   0.58000   0.67836       100\n",
            "            couch    0.51852   0.70000   0.59574       100\n",
            "            table    0.47727   0.63000   0.54310       100\n",
            "         wardrobe    0.70000   0.70000   0.70000       100\n",
            "              bee    0.39200   0.49000   0.43556       100\n",
            "           beetle    0.85897   0.67000   0.75281       100\n",
            "        butterfly    0.55303   0.73000   0.62931       100\n",
            "      caterpillar    0.82353   0.42000   0.55629       100\n",
            "        cockroach    0.97368   0.74000   0.84091       100\n",
            "             bear    0.58586   0.58000   0.58291       100\n",
            "          leopard    0.83333   0.85000   0.84158       100\n",
            "             lion    0.48235   0.82000   0.60741       100\n",
            "            tiger    0.67708   0.65000   0.66327       100\n",
            "             wolf    0.59184   0.29000   0.38926       100\n",
            "           bridge    0.46774   0.58000   0.51786       100\n",
            "           castle    0.51087   0.47000   0.48958       100\n",
            "            house    0.49032   0.76000   0.59608       100\n",
            "             road    0.82883   0.92000   0.87204       100\n",
            "       skyscraper    0.77320   0.75000   0.76142       100\n",
            "            cloud    0.65909   0.29000   0.40278       100\n",
            "           forest    0.52778   0.76000   0.62295       100\n",
            "         mountain    0.58491   0.62000   0.60194       100\n",
            "            plain    0.84906   0.90000   0.87379       100\n",
            "              sea    0.74528   0.79000   0.76699       100\n",
            "            camel    0.28481   0.45000   0.34884       100\n",
            "           cattle    0.90426   0.85000   0.87629       100\n",
            "       chimpanzee    0.78824   0.67000   0.72432       100\n",
            "         elephant    0.85556   0.77000   0.81053       100\n",
            "         kangaroo    0.57407   0.62000   0.59615       100\n",
            "              fox    0.85106   0.80000   0.82474       100\n",
            "        porcupine    0.73684   0.70000   0.71795       100\n",
            "           possum    0.69388   0.68000   0.68687       100\n",
            "          raccoon    0.56923   0.74000   0.64348       100\n",
            "            skunk    0.63235   0.43000   0.51190       100\n",
            "             crab    0.54545   0.54000   0.54271       100\n",
            "          lobster    0.67619   0.71000   0.69268       100\n",
            "            snail    0.70423   0.50000   0.58480       100\n",
            "           spider    0.90196   0.92000   0.91089       100\n",
            "             worm    0.73394   0.80000   0.76555       100\n",
            "             baby    0.68868   0.73000   0.70874       100\n",
            "              boy    0.68750   0.77000   0.72642       100\n",
            "             girl    0.25352   0.36000   0.29752       100\n",
            "              man    0.51261   0.61000   0.55708       100\n",
            "            woman    0.52577   0.51000   0.51777       100\n",
            "        crocodile    0.80000   0.88000   0.83810       100\n",
            "         dinosaur    0.85437   0.88000   0.86700       100\n",
            "           lizard    0.59821   0.67000   0.63208       100\n",
            "            snake    0.73684   0.56000   0.63636       100\n",
            "           turtle    0.74737   0.71000   0.72821       100\n",
            "          hamster    0.59091   0.39000   0.46988       100\n",
            "            mouse    0.61667   0.74000   0.67273       100\n",
            "           rabbit    0.81308   0.87000   0.84058       100\n",
            "            shrew    0.67442   0.58000   0.62366       100\n",
            "         squirrel    0.63333   0.57000   0.60000       100\n",
            "            maple    0.92958   0.66000   0.77193       100\n",
            "              oak    0.84211   0.64000   0.72727       100\n",
            "             palm    0.67327   0.68000   0.67662       100\n",
            "             pine    0.86207   0.75000   0.80214       100\n",
            "           willow    0.67521   0.79000   0.72811       100\n",
            "          bicycle    0.62069   0.72000   0.66667       100\n",
            "              bus    0.60145   0.83000   0.69748       100\n",
            "       motorcycle    0.51261   0.61000   0.55708       100\n",
            "     pickup truck    0.41447   0.63000   0.50000       100\n",
            "            train    0.81818   0.90000   0.85714       100\n",
            "       lawn-mower    0.70408   0.69000   0.69697       100\n",
            "           rocket    0.54839   0.51000   0.52850       100\n",
            "        streetcar    0.77632   0.59000   0.67045       100\n",
            "             tank    0.50617   0.41000   0.45304       100\n",
            "          tractor    0.81707   0.67000   0.73626       100\n",
            "\n",
            "         accuracy                        0.66150     10000\n",
            "        macro avg    0.68531   0.66150   0.66362     10000\n",
            "     weighted avg    0.68531   0.66150   0.66362     10000\n",
            "\n"
          ]
        }
      ]
    }
  ]
}