CNN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from tensorflow import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D\n",
    "from tensorflow.keras.applications import DenseNet121\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load the MNIST Dataset\n",
    "\n",
    "Function: [***Datasets***](https://keras.io/api/datasets/) in Keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train.shape = (50000, 28, 28, 1)\n",
      "y_train.shape = (50000,)\n",
      "x_val.shape = (10000, 28, 28, 1)\n",
      "y_val.shape = (10000,)\n",
      "x_test.shape = (10000, 28, 28, 1)\n",
      "y_test.shape = (10000,)\n",
      "\n",
      "This is one of the handwriting digits of \u001b[1m2:\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "(x_train, y_train),(x_test, y_test) = keras.datasets.mnist.load_data()\n",
    "x_train, x_test = x_train / 255.0, x_test / 255.0  \n",
    "\n",
    "x_train = x_train.reshape(-1,28,28,1)\n",
    "x_test = x_test.reshape(-1,28,28,1)\n",
    "\n",
    "# Split the data into training and testing datasets\n",
    "x_train, x_val, y_train, y_val = train_test_split(\n",
    "\tx_train, y_train, test_size = 1/6)\n",
    "\n",
    "# Print the dataset sizes\n",
    "print('x_train.shape =', x_train.shape)\n",
    "print('y_train.shape =', y_train.shape)\n",
    "print('x_val.shape =', x_val.shape)\n",
    "print('y_val.shape =', y_val.shape)\n",
    "print('x_test.shape =', x_test.shape)\n",
    "print('y_test.shape =', y_test.shape)\n",
    "\n",
    "# Pick a random integer/digit within the training set\n",
    "random_digit = np.random.randint(0, len(y_train)-1)\n",
    "\n",
    "# Plot the image of this digit for visualization and comparison after training\n",
    "print(\"\\nThis is one of the handwriting digits of \\033[1m\"+str(y_train[random_digit])+\":\\n\")\n",
    "plt.imshow(x_train[random_digit])\n",
    "plt.show()\n",
    "\n",
    "print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define and Compile ConvNet, Simple ANN and Transfer Learning Models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Layers\n",
    "\n",
    "Layers  Class: [***Layers***](https://keras.io/api/layers/) in Keras\n",
    "\n",
    "Flatten Layer: [***Flatten layer***](https://keras.io/api/layers/reshaping_layers/flatten/) in Keras\n",
    "\n",
    "Dense Layer: [***Dense layer***](https://keras.io/api/layers/core_layers/dense/) in Keras\n",
    "\n",
    "Conv2D Layer: [***Conv2D layer***](https://keras.io/api/layers/convolution_layers/convolution2d/) in Keras\n",
    "\n",
    "#### Metrics\n",
    "Metrics arguments: [***Metrics***](https://keras.io/api/metrics/) in Keras\n",
    "\n",
    "Accuracy arguments: [***Accuracy***](https://keras.io/api/metrics/accuracy_metrics/#accuracy-class) in Keras\n",
    "\n",
    "#### Losses\n",
    "\n",
    "Losses class: [***Losses***](https://keras.io/api/losses/) in Keras\n",
    "\n",
    "Sparse Categorical Crossentropy class: [***SparseCategoricalCrossentropy***](https://keras.io/api/losses/probabilistic_losses/#sparsecategoricalcrossentropy-class) in Keras\n",
    "\n",
    "#### Optimizers\n",
    "\n",
    "Optimizers arguments: [***Optimizers***](https://keras.io/api/optimizers/) in Keras\n",
    "\n",
    "Adam Class: [***Adam***](https://keras.io/api/models/model_training_apis/) in Keras\n",
    "\n",
    "#### Layer Activation\n",
    "\n",
    "Layer Activation: [***Layer activation functions***](https://keras.io/api/layers/activations/) in Keras\n",
    "\n",
    "ReLU function: [***ReLU***](https://keras.io/api/layers/activations/) in Keras\n",
    "\n",
    "Softmax function: [***Softmax***](https://keras.io/api/layers/activations/) in Keras\n",
    "\n",
    "\n",
    "#### Other\n",
    "Sequential Class: [***Sequential***](https://keras.io/api/models/sequential/) in Keras\n",
    "\n",
    "Summary Method: [***Summary***](https://keras.io/api/models/model/) in Keras\n",
    "\n",
    "Compile Method: [***Compile***](https://keras.io/api/models/model_training_apis/) in Keras\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define a function to generate and complie the ConvNet\n",
    "def make_conv_model():\n",
    "    model = Sequential()\n",
    "    \n",
    "    model.add(Conv2D(filters = 10, kernel_size = (5,5),padding = 'Same', \n",
    "                activation ='relu', input_shape = (28,28,1)))\n",
    "    model.add(Conv2D(filters = 10, kernel_size = (5,5),padding = 'Same', \n",
    "                activation ='relu'))\n",
    "    model.add(MaxPool2D(pool_size=(2,2)))\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(64, activation = \"relu\"))\n",
    "    model.add(Dropout(0.2))\n",
    "    model.add(Dense(10, activation = \"softmax\"))\n",
    "    \n",
    "    model.compile(optimizer=keras.optimizers.Adam(),\n",
    "                loss='sparse_categorical_crossentropy',\n",
    "                metrics=['accuracy'] ) \n",
    "    \n",
    "    model.summary()\n",
    "    return model\n",
    "\n",
    "# Define a function to generate and complie the ANN\n",
    "def make_dense_model():\n",
    "    model = Sequential()\n",
    "    \n",
    "    model.add(Flatten(input_shape = (28,28,1)))\n",
    "    model.add(Dense(512, activation = \"relu\"))\n",
    "    model.add(Dropout(0.2))\n",
    "    model.add(Dense(256, activation = \"relu\"))\n",
    "    model.add(Dropout(0.2))\n",
    "    model.add(Dense(10, activation = \"softmax\"))\n",
    "\n",
    "    model.compile(optimizer=keras.optimizers.Adam(),\n",
    "                loss='sparse_categorical_crossentropy',\n",
    "                metrics=['accuracy']  ) \n",
    "    \n",
    "    model.summary()\n",
    "    return model\n",
    "\n",
    "# Define a function to generate a transfer learning model from DenseNet121 model\n",
    "def make_transfer_learning_model():\n",
    "    model = Sequential()\n",
    "\n",
    "    model.add(keras.applications.DenseNet121(weights='imagenet',  # Load weights pre-trained on ImageNet.\n",
    "                                      include_top=False,          # Do not include the ImageNet classifier at the top.\n",
    "                                     input_shape = (32,32,3)))\n",
    "    #model.add(keras.layers.Flatten())\n",
    "    model.add(keras.layers.GlobalMaxPool2D())\n",
    "    #model.add(keras.layers.GlobalAveragePooling2D())\n",
    "    model.add(keras.layers.Dense(128, activation='relu'))\n",
    "    model.add(keras.layers.Dropout(0.2))  # Regularize with dropout\n",
    "    model.add(keras.layers.Dense(10, activation='softmax'))\n",
    "\n",
    "    for layer in model.layers[:-3]:\n",
    "        layer.trainable =  False\n",
    " \n",
    "    model.compile(optimizer=keras.optimizers.Adam(),\n",
    "                  loss='sparse_categorical_crossentropy',\n",
    "                  metrics=['accuracy'] ) \n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "    return model\n",
    "\n",
    "# Define the parameters for training the models\n",
    "epochs = 30\n",
    "batch_size = 256"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define and Train ConvNet and Simple ANN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fit Method: [***Fit***](https://keras.io/api/models/model_training_apis/) in Keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 28, 28, 10)        260       \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 28, 28, 10)        2510      \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 14, 14, 10)        0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 1960)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 64)                125504    \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 128,924\n",
      "Trainable params: 128,924\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "196/196 [==============================] - 36s 183ms/step - loss: 0.4871 - accuracy: 0.8525 - val_loss: 0.1491 - val_accuracy: 0.9558\n",
      "Epoch 2/30\n",
      "196/196 [==============================] - 38s 193ms/step - loss: 0.1420 - accuracy: 0.9569 - val_loss: 0.0904 - val_accuracy: 0.9721\n",
      "Epoch 3/30\n",
      "196/196 [==============================] - 29s 147ms/step - loss: 0.0995 - accuracy: 0.9697 - val_loss: 0.0807 - val_accuracy: 0.9748\n",
      "Epoch 4/30\n",
      "196/196 [==============================] - 28s 141ms/step - loss: 0.0781 - accuracy: 0.9764 - val_loss: 0.0670 - val_accuracy: 0.9802\n",
      "Epoch 5/30\n",
      "196/196 [==============================] - 29s 147ms/step - loss: 0.0649 - accuracy: 0.9798 - val_loss: 0.0571 - val_accuracy: 0.9833\n",
      "Epoch 6/30\n",
      "196/196 [==============================] - 29s 150ms/step - loss: 0.0551 - accuracy: 0.9830 - val_loss: 0.0650 - val_accuracy: 0.9810\n",
      "Epoch 7/30\n",
      "196/196 [==============================] - 30s 151ms/step - loss: 0.0466 - accuracy: 0.9852 - val_loss: 0.0522 - val_accuracy: 0.9852\n",
      "Epoch 8/30\n",
      "196/196 [==============================] - 33s 167ms/step - loss: 0.0405 - accuracy: 0.9873 - val_loss: 0.0570 - val_accuracy: 0.9841\n",
      "Epoch 9/30\n",
      "196/196 [==============================] - 35s 178ms/step - loss: 0.0383 - accuracy: 0.9878 - val_loss: 0.0471 - val_accuracy: 0.9866\n",
      "Epoch 10/30\n",
      "196/196 [==============================] - 31s 160ms/step - loss: 0.0314 - accuracy: 0.9900 - val_loss: 0.0495 - val_accuracy: 0.9864\n",
      "Epoch 11/30\n",
      "196/196 [==============================] - 29s 150ms/step - loss: 0.0294 - accuracy: 0.9903 - val_loss: 0.0582 - val_accuracy: 0.9849\n",
      "Epoch 12/30\n",
      "196/196 [==============================] - 31s 156ms/step - loss: 0.0277 - accuracy: 0.9911 - val_loss: 0.0517 - val_accuracy: 0.9865\n",
      "Epoch 13/30\n",
      "196/196 [==============================] - 33s 169ms/step - loss: 0.0251 - accuracy: 0.9915 - val_loss: 0.0569 - val_accuracy: 0.9863\n",
      "Epoch 14/30\n",
      "196/196 [==============================] - 33s 167ms/step - loss: 0.0239 - accuracy: 0.9923 - val_loss: 0.0533 - val_accuracy: 0.9859\n",
      "Epoch 15/30\n",
      "196/196 [==============================] - 30s 153ms/step - loss: 0.0216 - accuracy: 0.9926 - val_loss: 0.0510 - val_accuracy: 0.9876\n",
      "Epoch 16/30\n",
      "196/196 [==============================] - 29s 149ms/step - loss: 0.0193 - accuracy: 0.9935 - val_loss: 0.0498 - val_accuracy: 0.9874\n",
      "Epoch 17/30\n",
      "196/196 [==============================] - 30s 154ms/step - loss: 0.0188 - accuracy: 0.9935 - val_loss: 0.0504 - val_accuracy: 0.9878\n",
      "Epoch 18/30\n",
      "196/196 [==============================] - 34s 175ms/step - loss: 0.0183 - accuracy: 0.9939 - val_loss: 0.0517 - val_accuracy: 0.9873\n",
      "Epoch 19/30\n",
      "196/196 [==============================] - 31s 160ms/step - loss: 0.0170 - accuracy: 0.9944 - val_loss: 0.0534 - val_accuracy: 0.9854\n",
      "Epoch 20/30\n",
      "196/196 [==============================] - 30s 156ms/step - loss: 0.0164 - accuracy: 0.9943 - val_loss: 0.0537 - val_accuracy: 0.9875\n",
      "Epoch 21/30\n",
      "196/196 [==============================] - 30s 151ms/step - loss: 0.0160 - accuracy: 0.9944 - val_loss: 0.0537 - val_accuracy: 0.9880\n",
      "Epoch 22/30\n",
      "196/196 [==============================] - 31s 160ms/step - loss: 0.0152 - accuracy: 0.9946 - val_loss: 0.0576 - val_accuracy: 0.9870\n",
      "Epoch 23/30\n",
      "196/196 [==============================] - 32s 162ms/step - loss: 0.0132 - accuracy: 0.9951 - val_loss: 0.0537 - val_accuracy: 0.9873\n",
      "Epoch 24/30\n",
      "196/196 [==============================] - 30s 151ms/step - loss: 0.0129 - accuracy: 0.9956 - val_loss: 0.0583 - val_accuracy: 0.9876\n",
      "Epoch 25/30\n",
      "196/196 [==============================] - 30s 152ms/step - loss: 0.0127 - accuracy: 0.9956 - val_loss: 0.0597 - val_accuracy: 0.9876\n",
      "Epoch 26/30\n",
      "196/196 [==============================] - 33s 167ms/step - loss: 0.0135 - accuracy: 0.9954 - val_loss: 0.0553 - val_accuracy: 0.9880\n",
      "Epoch 27/30\n",
      "196/196 [==============================] - 33s 167ms/step - loss: 0.0118 - accuracy: 0.9965 - val_loss: 0.0641 - val_accuracy: 0.9871\n",
      "Epoch 28/30\n",
      "196/196 [==============================] - 31s 156ms/step - loss: 0.0114 - accuracy: 0.9962 - val_loss: 0.0603 - val_accuracy: 0.9867\n",
      "Epoch 29/30\n",
      "196/196 [==============================] - 30s 152ms/step - loss: 0.0103 - accuracy: 0.9963 - val_loss: 0.0654 - val_accuracy: 0.9876\n",
      "Epoch 30/30\n",
      "196/196 [==============================] - 32s 165ms/step - loss: 0.0095 - accuracy: 0.9968 - val_loss: 0.0607 - val_accuracy: 0.9880\n",
      "313/313 [==============================] - 2s 7ms/step - loss: 0.0424 - accuracy: 0.9900\n"
     ]
    }
   ],
   "source": [
    "simple_conv_model = make_conv_model()\n",
    "simple_conv_history = simple_conv_model.fit(x_train, y_train, batch_size=batch_size, epochs=epochs, validation_data=(x_val, y_val), verbose=1)\n",
    "simple_conv_test_loss, simple_conv_test_acc = simple_conv_model.evaluate(x_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fit Method: [***Fit***](https://keras.io/api/models/model_training_apis/) in Keras\n",
    "\n",
    "Evaluate Method: [***Evaluate***](https://keras.io/api/models/model_training_apis/) in Keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten_1 (Flatten)          (None, 784)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 512)               401920    \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 10)                2570      \n",
      "=================================================================\n",
      "Total params: 535,818\n",
      "Trainable params: 535,818\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "196/196 [==============================] - 2s 12ms/step - loss: 0.3489 - accuracy: 0.8961 - val_loss: 0.1474 - val_accuracy: 0.9560\n",
      "Epoch 2/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.1304 - accuracy: 0.9614 - val_loss: 0.1062 - val_accuracy: 0.9690\n",
      "Epoch 3/30\n",
      "196/196 [==============================] - 2s 12ms/step - loss: 0.0887 - accuracy: 0.9733 - val_loss: 0.0890 - val_accuracy: 0.9725\n",
      "Epoch 4/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0690 - accuracy: 0.9792 - val_loss: 0.0814 - val_accuracy: 0.9753\n",
      "Epoch 5/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0538 - accuracy: 0.9834 - val_loss: 0.0759 - val_accuracy: 0.9775\n",
      "Epoch 6/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0415 - accuracy: 0.9869 - val_loss: 0.0730 - val_accuracy: 0.9776\n",
      "Epoch 7/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0357 - accuracy: 0.9884 - val_loss: 0.0739 - val_accuracy: 0.9795\n",
      "Epoch 8/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0309 - accuracy: 0.9900 - val_loss: 0.0774 - val_accuracy: 0.9772\n",
      "Epoch 9/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0262 - accuracy: 0.9915 - val_loss: 0.0762 - val_accuracy: 0.9804\n",
      "Epoch 10/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0241 - accuracy: 0.9919 - val_loss: 0.0714 - val_accuracy: 0.9798\n",
      "Epoch 11/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0208 - accuracy: 0.9931 - val_loss: 0.0903 - val_accuracy: 0.9766\n",
      "Epoch 12/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0195 - accuracy: 0.9932 - val_loss: 0.0842 - val_accuracy: 0.9791\n",
      "Epoch 13/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0184 - accuracy: 0.9939 - val_loss: 0.0788 - val_accuracy: 0.9798\n",
      "Epoch 14/30\n",
      "196/196 [==============================] - 2s 11ms/step - loss: 0.0166 - accuracy: 0.9942 - val_loss: 0.0804 - val_accuracy: 0.9786\n",
      "Epoch 15/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0158 - accuracy: 0.9944 - val_loss: 0.0803 - val_accuracy: 0.9794\n",
      "Epoch 16/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0156 - accuracy: 0.9945 - val_loss: 0.0864 - val_accuracy: 0.9800\n",
      "Epoch 17/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0138 - accuracy: 0.9954 - val_loss: 0.0858 - val_accuracy: 0.9784\n",
      "Epoch 18/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0132 - accuracy: 0.9955 - val_loss: 0.0945 - val_accuracy: 0.9778\n",
      "Epoch 19/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0127 - accuracy: 0.9956 - val_loss: 0.0855 - val_accuracy: 0.9804\n",
      "Epoch 20/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0127 - accuracy: 0.9957 - val_loss: 0.0845 - val_accuracy: 0.9798\n",
      "Epoch 21/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0111 - accuracy: 0.9959 - val_loss: 0.0953 - val_accuracy: 0.9791\n",
      "Epoch 22/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0102 - accuracy: 0.9967 - val_loss: 0.0945 - val_accuracy: 0.9802\n",
      "Epoch 23/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0104 - accuracy: 0.9963 - val_loss: 0.0930 - val_accuracy: 0.9801\n",
      "Epoch 24/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0111 - accuracy: 0.9959 - val_loss: 0.1062 - val_accuracy: 0.9778\n",
      "Epoch 25/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0119 - accuracy: 0.9961 - val_loss: 0.1030 - val_accuracy: 0.9778\n",
      "Epoch 26/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0104 - accuracy: 0.9964 - val_loss: 0.0904 - val_accuracy: 0.9802\n",
      "Epoch 27/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0118 - accuracy: 0.9961 - val_loss: 0.1014 - val_accuracy: 0.9797\n",
      "Epoch 28/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0085 - accuracy: 0.9970 - val_loss: 0.0915 - val_accuracy: 0.9814\n",
      "Epoch 29/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0094 - accuracy: 0.9967 - val_loss: 0.1029 - val_accuracy: 0.9794\n",
      "Epoch 30/30\n",
      "196/196 [==============================] - 2s 10ms/step - loss: 0.0100 - accuracy: 0.9966 - val_loss: 0.1030 - val_accuracy: 0.9813\n",
      "313/313 [==============================] - 1s 2ms/step - loss: 0.0881 - accuracy: 0.9815\n"
     ]
    }
   ],
   "source": [
    "dense_model = make_dense_model()\n",
    "dense_history = dense_model.fit(x_train, y_train, batch_size=batch_size, epochs=epochs, validation_data=(x_val, y_val),verbose=1)\n",
    "dense_test_loss, dense_test_acc = dense_model.evaluate(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "densenet121 (Functional)     (None, 1, 1, 1024)        7037504   \n",
      "_________________________________________________________________\n",
      "global_max_pooling2d (Global (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 128)               131200    \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 7,169,994\n",
      "Trainable params: 132,490\n",
      "Non-trainable params: 7,037,504\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "196/196 [==============================] - 99s 506ms/step - loss: 0.6718 - accuracy: 0.7872 - val_loss: 0.3028 - val_accuracy: 0.9075\n",
      "Epoch 2/30\n",
      "196/196 [==============================] - 97s 497ms/step - loss: 0.2837 - accuracy: 0.9115 - val_loss: 0.2122 - val_accuracy: 0.9328\n",
      "Epoch 3/30\n",
      "196/196 [==============================] - 97s 497ms/step - loss: 0.2179 - accuracy: 0.9296 - val_loss: 0.1825 - val_accuracy: 0.9417\n",
      "Epoch 4/30\n",
      "196/196 [==============================] - 97s 497ms/step - loss: 0.1844 - accuracy: 0.9415 - val_loss: 0.1610 - val_accuracy: 0.9492\n",
      "Epoch 5/30\n",
      "196/196 [==============================] - 97s 493ms/step - loss: 0.1679 - accuracy: 0.9464 - val_loss: 0.1605 - val_accuracy: 0.9464\n",
      "Epoch 6/30\n",
      "196/196 [==============================] - 98s 500ms/step - loss: 0.1548 - accuracy: 0.9498 - val_loss: 0.1425 - val_accuracy: 0.9538\n",
      "Epoch 7/30\n",
      "196/196 [==============================] - 95s 485ms/step - loss: 0.1416 - accuracy: 0.9544 - val_loss: 0.1400 - val_accuracy: 0.9551\n",
      "Epoch 8/30\n",
      "196/196 [==============================] - 101s 518ms/step - loss: 0.1340 - accuracy: 0.9573 - val_loss: 0.1337 - val_accuracy: 0.9570\n",
      "Epoch 9/30\n",
      "196/196 [==============================] - 94s 479ms/step - loss: 0.1264 - accuracy: 0.9585 - val_loss: 0.1304 - val_accuracy: 0.9584\n",
      "Epoch 10/30\n",
      "196/196 [==============================] - 98s 501ms/step - loss: 0.1211 - accuracy: 0.9597 - val_loss: 0.1279 - val_accuracy: 0.9592\n",
      "Epoch 11/30\n",
      "196/196 [==============================] - 98s 501ms/step - loss: 0.1148 - accuracy: 0.9623 - val_loss: 0.1193 - val_accuracy: 0.9609\n",
      "Epoch 12/30\n",
      "196/196 [==============================] - 99s 505ms/step - loss: 0.1115 - accuracy: 0.9635 - val_loss: 0.1176 - val_accuracy: 0.9636\n",
      "Epoch 13/30\n",
      "196/196 [==============================] - 97s 496ms/step - loss: 0.1092 - accuracy: 0.9649 - val_loss: 0.1173 - val_accuracy: 0.9639\n",
      "Epoch 14/30\n",
      "196/196 [==============================] - 100s 509ms/step - loss: 0.1021 - accuracy: 0.9666 - val_loss: 0.1391 - val_accuracy: 0.9551\n",
      "Epoch 15/30\n",
      "196/196 [==============================] - 95s 483ms/step - loss: 0.0993 - accuracy: 0.9672 - val_loss: 0.1158 - val_accuracy: 0.9643\n",
      "Epoch 16/30\n",
      "196/196 [==============================] - 98s 501ms/step - loss: 0.0966 - accuracy: 0.9688 - val_loss: 0.1130 - val_accuracy: 0.9640\n",
      "Epoch 17/30\n",
      "196/196 [==============================] - 100s 512ms/step - loss: 0.0910 - accuracy: 0.9700 - val_loss: 0.1206 - val_accuracy: 0.9618\n",
      "Epoch 18/30\n",
      "196/196 [==============================] - 94s 480ms/step - loss: 0.0908 - accuracy: 0.9695 - val_loss: 0.1082 - val_accuracy: 0.9660\n",
      "Epoch 19/30\n",
      "196/196 [==============================] - 102s 519ms/step - loss: 0.0855 - accuracy: 0.9718 - val_loss: 0.1143 - val_accuracy: 0.9640\n",
      "Epoch 20/30\n",
      "196/196 [==============================] - 94s 478ms/step - loss: 0.0860 - accuracy: 0.9716 - val_loss: 0.1318 - val_accuracy: 0.9584\n",
      "Epoch 21/30\n",
      "196/196 [==============================] - 100s 509ms/step - loss: 0.0814 - accuracy: 0.9723 - val_loss: 0.1125 - val_accuracy: 0.9654\n",
      "Epoch 22/30\n",
      "196/196 [==============================] - 97s 494ms/step - loss: 0.0795 - accuracy: 0.9732 - val_loss: 0.1077 - val_accuracy: 0.9665\n",
      "Epoch 23/30\n",
      "196/196 [==============================] - 98s 501ms/step - loss: 0.0788 - accuracy: 0.9739 - val_loss: 0.1072 - val_accuracy: 0.9661\n",
      "Epoch 24/30\n",
      "196/196 [==============================] - 99s 507ms/step - loss: 0.0760 - accuracy: 0.9753 - val_loss: 0.1051 - val_accuracy: 0.9675\n",
      "Epoch 25/30\n",
      "196/196 [==============================] - 98s 502ms/step - loss: 0.0727 - accuracy: 0.9757 - val_loss: 0.1082 - val_accuracy: 0.9661\n",
      "Epoch 26/30\n",
      "196/196 [==============================] - 98s 500ms/step - loss: 0.0693 - accuracy: 0.9767 - val_loss: 0.1043 - val_accuracy: 0.9666\n",
      "Epoch 27/30\n",
      "196/196 [==============================] - 223s 1s/step - loss: 0.0694 - accuracy: 0.9773 - val_loss: 0.1103 - val_accuracy: 0.9672\n",
      "Epoch 28/30\n",
      "196/196 [==============================] - 223s 1s/step - loss: 0.0698 - accuracy: 0.9763 - val_loss: 0.1047 - val_accuracy: 0.9683\n",
      "Epoch 29/30\n",
      "196/196 [==============================] - 221s 1s/step - loss: 0.0662 - accuracy: 0.9781 - val_loss: 0.1056 - val_accuracy: 0.9682\n",
      "Epoch 30/30\n",
      "196/196 [==============================] - 132s 674ms/step - loss: 0.0656 - accuracy: 0.9778 - val_loss: 0.1138 - val_accuracy: 0.9661\n",
      "313/313 [==============================] - 20s 64ms/step - loss: 0.0952 - accuracy: 0.9671\n"
     ]
    }
   ],
   "source": [
    "# add padding to the images to make them as 32 x 32, because pre-trained deep learning models require at least 32 x 32 input\n",
    "x_train_exp = np.pad(x_train,((0,0),(2,2),(2,2), (0,0)), 'constant')\n",
    "x_test_exp  = np.pad(x_test, ((0,0),(2,2),(2,2), (0,0)), 'constant')\n",
    "x_val_exp   = np.pad(x_val,  ((0,0),(2,2),(2,2), (0,0)), 'constant')\n",
    "\n",
    "# create 3 channel image datasets, because pre-trained deep learning models require exactly 3 inputs channels\n",
    "# we give the same values to the RGB channels, so the output is in black colors\n",
    "x_train3 = np.concatenate((x_train_exp,x_train_exp,x_train_exp), axis = 3)\n",
    "x_test3  = np.concatenate((x_test_exp,x_test_exp,x_test_exp), axis = 3)\n",
    "x_val3   = np.concatenate((x_val_exp,x_val_exp,x_val_exp), axis = 3)\n",
    "\n",
    "#print(x_train3.shape, x_test3.shape, x_val3.shape)   \n",
    "plt.imshow(x_train3[random_digit])\n",
    "plt.show()\n",
    "\n",
    "#use the function to generate the transfer learning model\n",
    "transfer_learning_model = make_transfer_learning_model()\n",
    "\n",
    "'''\n",
    "# print out the initial loss and accuracy \"before\" training\n",
    "loss0, accuracy0 = transfer_learning_model.evaluate(x_val3, y_val)\n",
    "print(\"initial loss: {:.2f}\".format(loss0))\n",
    "print(\"initial accuracy: {:.2f}%\".format(accuracy0*100))\n",
    "'''\n",
    "\n",
    "transfer_learning_history = transfer_learning_model.fit(x_train3, y_train, batch_size=batch_size,\n",
    "                                epochs = epochs, validation_data = (x_val3, y_val), verbose=1)\n",
    "\n",
    "\n",
    "transfer_learning_test_loss, transfer_learning_test_acc = transfer_learning_model.evaluate(x_test3, y_test)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the accuracy curves of both models for training, evaluation and testing data\n",
    "plt.figure(figsize=(12,8))\n",
    "plt.plot(simple_conv_history.history[\"accuracy\"], \"g--\",label=\"simple conv training accuracy\")\n",
    "plt.plot(simple_conv_history.history[\"val_accuracy\"], \"g-\",label=\"simple conv val accuracy\")\n",
    "plt.axhline(simple_conv_test_acc,color=\"k\", linestyle=\"--\", label=\"simple conv test accuracy\")\n",
    "\n",
    "plt.plot(dense_history.history[\"accuracy\"], \"b--\",label=\"dense training accuracy\")\n",
    "plt.plot(dense_history.history[\"val_accuracy\"], \"b-\",label=\"dense val accuracy\")\n",
    "plt.axhline(dense_test_acc, color=\"k\", linestyle=\":\", label=\"dense test accuracy\")\n",
    "\n",
    "plt.plot(transfer_learning_history.history[\"accuracy\"], \"r--\",label=\"trans. learn training accuracy\")\n",
    "plt.plot(transfer_learning_history.history[\"val_accuracy\"], \"r-\",label=\"trans. learn  val accuracy\")\n",
    "plt.axhline(transfer_learning_test_acc, color=\"k\", linestyle=\"-.\", label=\"trans. learn  test accuracy\")\n",
    "\n",
    "plt.xlabel(\"epochs\", fontsize=15)\n",
    "plt.ylabel(\"accuracy\", fontsize=15)\n",
    "plt.ylim(0.80, 1.00)\n",
    "plt.legend()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Augmentation\n",
    "\n",
    "ImageDataGenerator function: [***ImageDataGenerator***](https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image/ImageDataGenerator) in Keras\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Perform data augmentation over the image datasets\n",
    "datagen = ImageDataGenerator(\n",
    "        # set input mean to 0 over the dataset\n",
    "        featurewise_center=False, \n",
    "         # set each sample mean to 0\n",
    "        samplewise_center=False, \n",
    "        # divide inputs by std of the dataset\n",
    "        featurewise_std_normalization=False, \n",
    "        # divide each input by its std\n",
    "        samplewise_std_normalization=False,\n",
    "        # apply ZCA whitening\n",
    "        zca_whitening=False,  \n",
    "        # randomly rotate images by degrees\n",
    "        rotation_range=10,  \n",
    "        # Randomly zoom image\n",
    "        zoom_range = 0.1,  \n",
    "        # randomly shift images horizontally \n",
    "        width_shift_range=0.1,  \n",
    "        # randomly shift images vertically \n",
    "        height_shift_range=0.1,  \n",
    "        # randomly flip images\n",
    "        horizontal_flip=False,  \n",
    "        # randomly flip images\n",
    "        vertical_flip=False)  \n",
    "\n",
    "datagen.fit(x_train.reshape(-1,28,28,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_2 (Conv2D)            (None, 28, 28, 10)        260       \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 28, 28, 10)        2510      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 14, 14, 10)        0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 1960)              0         \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 64)                125504    \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 128,924\n",
      "Trainable params: 128,924\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "195/195 [==============================] - 35s 180ms/step - loss: 0.6636 - accuracy: 0.7919 - val_loss: 0.1296 - val_accuracy: 0.9627\n",
      "Epoch 2/30\n",
      "195/195 [==============================] - 40s 203ms/step - loss: 0.2318 - accuracy: 0.9304 - val_loss: 0.0893 - val_accuracy: 0.9718\n",
      "Epoch 3/30\n",
      "195/195 [==============================] - 38s 194ms/step - loss: 0.1840 - accuracy: 0.9447 - val_loss: 0.0763 - val_accuracy: 0.9768\n",
      "Epoch 4/30\n",
      "195/195 [==============================] - 38s 196ms/step - loss: 0.1569 - accuracy: 0.9521 - val_loss: 0.0624 - val_accuracy: 0.9805\n",
      "Epoch 5/30\n",
      "195/195 [==============================] - 38s 196ms/step - loss: 0.1377 - accuracy: 0.9584 - val_loss: 0.0601 - val_accuracy: 0.9828\n",
      "Epoch 6/30\n",
      "195/195 [==============================] - 37s 189ms/step - loss: 0.1211 - accuracy: 0.9635 - val_loss: 0.0546 - val_accuracy: 0.9835\n",
      "Epoch 7/30\n",
      "195/195 [==============================] - 33s 168ms/step - loss: 0.1121 - accuracy: 0.9652 - val_loss: 0.0525 - val_accuracy: 0.9840\n",
      "Epoch 8/30\n",
      "195/195 [==============================] - 34s 172ms/step - loss: 0.1082 - accuracy: 0.9676 - val_loss: 0.0464 - val_accuracy: 0.9870\n",
      "Epoch 9/30\n",
      "195/195 [==============================] - 37s 191ms/step - loss: 0.1005 - accuracy: 0.9691 - val_loss: 0.0457 - val_accuracy: 0.9869\n",
      "Epoch 10/30\n",
      "195/195 [==============================] - 38s 195ms/step - loss: 0.0945 - accuracy: 0.9704 - val_loss: 0.0409 - val_accuracy: 0.9875\n",
      "Epoch 11/30\n",
      "195/195 [==============================] - 36s 187ms/step - loss: 0.0881 - accuracy: 0.9728 - val_loss: 0.0413 - val_accuracy: 0.9885\n",
      "Epoch 12/30\n",
      "195/195 [==============================] - 33s 171ms/step - loss: 0.0858 - accuracy: 0.9744 - val_loss: 0.0423 - val_accuracy: 0.9874\n",
      "Epoch 13/30\n",
      "195/195 [==============================] - 34s 172ms/step - loss: 0.0803 - accuracy: 0.9754 - val_loss: 0.0392 - val_accuracy: 0.9880\n",
      "Epoch 14/30\n",
      "195/195 [==============================] - 36s 185ms/step - loss: 0.0783 - accuracy: 0.9757 - val_loss: 0.0406 - val_accuracy: 0.9883\n",
      "Epoch 15/30\n",
      "195/195 [==============================] - 37s 190ms/step - loss: 0.0759 - accuracy: 0.9770 - val_loss: 0.0430 - val_accuracy: 0.9875\n",
      "Epoch 16/30\n",
      "195/195 [==============================] - 35s 177ms/step - loss: 0.0719 - accuracy: 0.9784 - val_loss: 0.0352 - val_accuracy: 0.9897\n",
      "Epoch 17/30\n",
      "195/195 [==============================] - 34s 174ms/step - loss: 0.0681 - accuracy: 0.9787 - val_loss: 0.0353 - val_accuracy: 0.9894\n",
      "Epoch 18/30\n",
      "195/195 [==============================] - 38s 195ms/step - loss: 0.0678 - accuracy: 0.9797 - val_loss: 0.0382 - val_accuracy: 0.9884\n",
      "Epoch 19/30\n",
      "195/195 [==============================] - 35s 181ms/step - loss: 0.0645 - accuracy: 0.9810 - val_loss: 0.0345 - val_accuracy: 0.9904\n",
      "Epoch 20/30\n",
      "195/195 [==============================] - 33s 167ms/step - loss: 0.0662 - accuracy: 0.9793 - val_loss: 0.0388 - val_accuracy: 0.9887\n",
      "Epoch 21/30\n",
      "195/195 [==============================] - 33s 171ms/step - loss: 0.0614 - accuracy: 0.9810 - val_loss: 0.0368 - val_accuracy: 0.9899\n",
      "Epoch 22/30\n",
      "195/195 [==============================] - 37s 191ms/step - loss: 0.0637 - accuracy: 0.9799 - val_loss: 0.0418 - val_accuracy: 0.9875\n",
      "Epoch 23/30\n",
      "195/195 [==============================] - 37s 189ms/step - loss: 0.0610 - accuracy: 0.9813 - val_loss: 0.0346 - val_accuracy: 0.9909\n",
      "Epoch 24/30\n",
      "195/195 [==============================] - 33s 171ms/step - loss: 0.0584 - accuracy: 0.9825 - val_loss: 0.0359 - val_accuracy: 0.9901\n",
      "Epoch 25/30\n",
      "195/195 [==============================] - 33s 170ms/step - loss: 0.0585 - accuracy: 0.9817 - val_loss: 0.0342 - val_accuracy: 0.9906\n",
      "Epoch 26/30\n",
      "195/195 [==============================] - 37s 189ms/step - loss: 0.0557 - accuracy: 0.9829 - val_loss: 0.0334 - val_accuracy: 0.9905\n",
      "Epoch 27/30\n",
      "195/195 [==============================] - 37s 191ms/step - loss: 0.0525 - accuracy: 0.9843 - val_loss: 0.0346 - val_accuracy: 0.9909\n",
      "Epoch 28/30\n",
      "195/195 [==============================] - 33s 169ms/step - loss: 0.0565 - accuracy: 0.9823 - val_loss: 0.0361 - val_accuracy: 0.9907\n",
      "Epoch 29/30\n",
      "195/195 [==============================] - 35s 178ms/step - loss: 0.0560 - accuracy: 0.9827 - val_loss: 0.0319 - val_accuracy: 0.9915\n",
      "Epoch 30/30\n",
      "195/195 [==============================] - 38s 194ms/step - loss: 0.0536 - accuracy: 0.9836 - val_loss: 0.0317 - val_accuracy: 0.9916\n",
      "313/313 [==============================] - 2s 6ms/step - loss: 0.0208 - accuracy: 0.9929\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# use the function to generate the ConvNet model\n",
    "simple_conv_model = make_conv_model()\n",
    "\n",
    "simple_conv_aug_history = simple_conv_model.fit(\n",
    "    datagen.flow(x_train,y_train, batch_size=batch_size),\n",
    "    epochs = epochs, validation_data = (x_val,y_val),\n",
    "    steps_per_epoch=x_train.shape[0]//batch_size, verbose=1)\n",
    "\n",
    "simple_conv_aug_test_loss, simple_conv_aug_test_acc = simple_conv_model.evaluate(x_test, y_test)\n",
    "\n",
    "plt.plot(simple_conv_aug_history.history[\"accuracy\"], \"g--\",label=\"simple conv aug training accuracy\")\n",
    "plt.plot(simple_conv_aug_history.history[\"val_accuracy\"], \"g-\",label=\"simple conv aug val accuracy\")\n",
    "plt.axhline(simple_conv_aug_test_acc,color=\"k\", linestyle=\"--\", label=\"simple conv aug test accuracy\")\n",
    "plt.axhline(simple_conv_test_acc,color=\"r\", linestyle=\"--\", label=\"simple conv test accuracy\")\n",
    "plt.xlabel(\"epochs\", fontsize=15)\n",
    "plt.ylabel(\"accuracy\", fontsize=15)\n",
    "plt.ylim(0.80, 1.005)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_4\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "densenet121 (Functional)     (None, 1, 1, 1024)        7037504   \n",
      "_________________________________________________________________\n",
      "global_max_pooling2d_1 (Glob (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 128)               131200    \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_10 (Dense)             (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 7,169,994\n",
      "Trainable params: 132,490\n",
      "Non-trainable params: 7,037,504\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "196/196 [==============================] - 107s 546ms/step - loss: 0.7741 - accuracy: 0.7502 - val_loss: 0.4072 - val_accuracy: 0.8715\n",
      "Epoch 2/30\n",
      "196/196 [==============================] - 111s 565ms/step - loss: 0.3841 - accuracy: 0.8767 - val_loss: 0.3339 - val_accuracy: 0.8912\n",
      "Epoch 3/30\n",
      "196/196 [==============================] - 107s 547ms/step - loss: 0.3117 - accuracy: 0.8990 - val_loss: 0.2739 - val_accuracy: 0.9105\n",
      "Epoch 4/30\n",
      "196/196 [==============================] - 109s 558ms/step - loss: 0.2792 - accuracy: 0.9078 - val_loss: 0.2431 - val_accuracy: 0.9215\n",
      "Epoch 5/30\n",
      "196/196 [==============================] - 108s 551ms/step - loss: 0.2537 - accuracy: 0.9181 - val_loss: 0.2248 - val_accuracy: 0.9265\n",
      "Epoch 6/30\n",
      "196/196 [==============================] - 107s 547ms/step - loss: 0.2447 - accuracy: 0.9196 - val_loss: 0.2392 - val_accuracy: 0.9192\n",
      "Epoch 7/30\n",
      "196/196 [==============================] - 107s 547ms/step - loss: 0.2274 - accuracy: 0.9252 - val_loss: 0.2121 - val_accuracy: 0.9292\n",
      "Epoch 8/30\n",
      "196/196 [==============================] - 110s 563ms/step - loss: 0.2171 - accuracy: 0.9299 - val_loss: 0.2145 - val_accuracy: 0.9273\n",
      "Epoch 9/30\n",
      "196/196 [==============================] - 106s 541ms/step - loss: 0.2105 - accuracy: 0.9318 - val_loss: 0.2048 - val_accuracy: 0.9312\n",
      "Epoch 10/30\n",
      "196/196 [==============================] - 107s 545ms/step - loss: 0.2078 - accuracy: 0.9319 - val_loss: 0.2316 - val_accuracy: 0.9212\n",
      "Epoch 11/30\n",
      "196/196 [==============================] - 108s 549ms/step - loss: 0.1969 - accuracy: 0.9348 - val_loss: 0.2002 - val_accuracy: 0.9328\n",
      "Epoch 12/30\n",
      "196/196 [==============================] - 108s 553ms/step - loss: 0.1910 - accuracy: 0.9361 - val_loss: 0.1853 - val_accuracy: 0.9366\n",
      "Epoch 13/30\n",
      "196/196 [==============================] - 107s 548ms/step - loss: 0.1860 - accuracy: 0.9382 - val_loss: 0.1918 - val_accuracy: 0.9338\n",
      "Epoch 14/30\n",
      "196/196 [==============================] - 109s 557ms/step - loss: 0.1904 - accuracy: 0.9372 - val_loss: 0.1832 - val_accuracy: 0.9394\n",
      "Epoch 15/30\n",
      "196/196 [==============================] - 107s 543ms/step - loss: 0.1823 - accuracy: 0.9402 - val_loss: 0.1822 - val_accuracy: 0.9406\n",
      "Epoch 16/30\n",
      "196/196 [==============================] - 108s 551ms/step - loss: 0.1749 - accuracy: 0.9428 - val_loss: 0.1939 - val_accuracy: 0.9353\n",
      "Epoch 17/30\n",
      "196/196 [==============================] - 110s 563ms/step - loss: 0.1783 - accuracy: 0.9413 - val_loss: 0.1812 - val_accuracy: 0.9382\n",
      "Epoch 18/30\n",
      "196/196 [==============================] - 105s 536ms/step - loss: 0.1707 - accuracy: 0.9444 - val_loss: 0.1534 - val_accuracy: 0.9485\n",
      "Epoch 19/30\n",
      "196/196 [==============================] - 110s 563ms/step - loss: 0.1688 - accuracy: 0.9442 - val_loss: 0.1914 - val_accuracy: 0.9352\n",
      "Epoch 20/30\n",
      "196/196 [==============================] - 105s 536ms/step - loss: 0.1690 - accuracy: 0.9441 - val_loss: 0.1746 - val_accuracy: 0.9432\n",
      "Epoch 21/30\n",
      "196/196 [==============================] - 108s 550ms/step - loss: 0.1666 - accuracy: 0.9451 - val_loss: 0.1779 - val_accuracy: 0.9412\n",
      "Epoch 22/30\n",
      "196/196 [==============================] - 110s 560ms/step - loss: 0.1691 - accuracy: 0.9440 - val_loss: 0.1800 - val_accuracy: 0.9406\n",
      "Epoch 23/30\n",
      "196/196 [==============================] - 108s 552ms/step - loss: 0.1664 - accuracy: 0.9449 - val_loss: 0.1717 - val_accuracy: 0.9438\n",
      "Epoch 24/30\n",
      "196/196 [==============================] - 109s 558ms/step - loss: 0.1627 - accuracy: 0.9455 - val_loss: 0.1769 - val_accuracy: 0.9412\n",
      "Epoch 25/30\n",
      "196/196 [==============================] - 104s 531ms/step - loss: 0.1636 - accuracy: 0.9454 - val_loss: 0.1708 - val_accuracy: 0.9421\n",
      "Epoch 26/30\n",
      "196/196 [==============================] - 109s 554ms/step - loss: 0.1640 - accuracy: 0.9449 - val_loss: 0.1630 - val_accuracy: 0.9450\n",
      "Epoch 27/30\n",
      "196/196 [==============================] - 112s 569ms/step - loss: 0.1569 - accuracy: 0.9474 - val_loss: 0.1663 - val_accuracy: 0.9450\n",
      "Epoch 28/30\n",
      "196/196 [==============================] - 108s 549ms/step - loss: 0.1594 - accuracy: 0.9487 - val_loss: 0.1832 - val_accuracy: 0.9395\n",
      "Epoch 29/30\n",
      "196/196 [==============================] - 104s 531ms/step - loss: 0.1555 - accuracy: 0.9491 - val_loss: 0.1696 - val_accuracy: 0.9445\n",
      "Epoch 30/30\n",
      "196/196 [==============================] - 109s 557ms/step - loss: 0.1532 - accuracy: 0.9495 - val_loss: 0.1560 - val_accuracy: 0.9473\n",
      "313/313 [==============================] - 21s 66ms/step - loss: 0.1362 - accuracy: 0.9545\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# use the function to generate the transfer learning model\n",
    "transfer_learning_model = make_transfer_learning_model()\n",
    "\n",
    "transfer_learning_aug_history = transfer_learning_model.fit(\n",
    "                                datagen.flow(x_train3,y_train, batch_size=batch_size),\n",
    "                                epochs = epochs, validation_data = (x_val3, y_val), verbose=1)\n",
    "\n",
    "transfer_learning_aug_test_loss, transfer_learning_aug_test_acc = transfer_learning_model.evaluate(x_test3, y_test)\n",
    "\n",
    "\n",
    "plt.plot(transfer_learning_aug_history.history[\"accuracy\"], \"g--\",label=\"trans. learn aug training accuracy\")\n",
    "plt.plot(transfer_learning_aug_history.history[\"val_accuracy\"], \"g-\",label=\"trans. learn aug val accuracy\")\n",
    "plt.axhline(transfer_learning_aug_test_acc,color=\"k\", linestyle=\"--\", label=\"trans. learn aug test accuracy\")\n",
    "plt.axhline(transfer_learning_test_acc,color=\"r\", linestyle=\"--\", label=\"trans. learn test accuracy\")\n",
    "plt.xlabel(\"epochs\", fontsize=15)\n",
    "plt.ylabel(\"accuracy\", fontsize=15)\n",
    "plt.ylim(0.80, 1.005)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}