02_03_deep_learning_conv_neural_network.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Your first convolutional neural network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "from tensorflow.keras.layers import Input, Flatten, Dense, Conv2D, BatchNormalization, LeakyReLU, Dropout, Activation\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "import tensorflow.keras.backend as K \n",
    "\n",
    "from tensorflow.keras.datasets import cifar10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "NUM_CLASSES = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_train, y_train), (x_test, y_test) = cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = x_train.astype('float32') / 255.0\n",
    "x_test = x_test.astype('float32') / 255.0\n",
    "\n",
    "y_train = to_categorical(y_train, NUM_CLASSES)\n",
    "y_test = to_categorical(y_test, NUM_CLASSES)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train[54, 12, 13, 1] "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_layer = Input(shape=(32,32,3))\n",
    "\n",
    "conv_layer_1 = Conv2D(\n",
    "    filters = 10\n",
    "    , kernel_size = (4,4)\n",
    "    , strides = 2\n",
    "    , padding = 'same'\n",
    "    )(input_layer)\n",
    "\n",
    "conv_layer_2 = Conv2D(\n",
    "    filters = 20\n",
    "    , kernel_size = (3,3)\n",
    "    , strides = 2\n",
    "    , padding = 'same'\n",
    "    )(conv_layer_1)\n",
    "\n",
    "flatten_layer = Flatten()(conv_layer_2)\n",
    "\n",
    "output_layer = Dense(units=10, activation = 'softmax')(flatten_layer)\n",
    "\n",
    "model = Model(input_layer, output_layer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_layer = Input((32,32,3))\n",
    "\n",
    "x = Conv2D(filters = 32, kernel_size = 3, strides = 1, padding = 'same')(input_layer)\n",
    "x = BatchNormalization()(x)\n",
    "x = LeakyReLU()(x)\n",
    "\n",
    "\n",
    "x = Conv2D(filters = 32, kernel_size = 3, strides = 2, padding = 'same')(x)\n",
    "x = BatchNormalization()(x)\n",
    "x = LeakyReLU()(x)\n",
    "\n",
    "\n",
    "x = Conv2D(filters = 64, kernel_size = 3, strides = 1, padding = 'same')(x)\n",
    "x = BatchNormalization()(x)\n",
    "x = LeakyReLU()(x)\n",
    "\n",
    "\n",
    "x = Conv2D(filters = 64, kernel_size = 3, strides = 2, padding = 'same')(x)\n",
    "x = BatchNormalization()(x)\n",
    "x = LeakyReLU()(x)\n",
    "\n",
    "\n",
    "x = Flatten()(x)\n",
    "\n",
    "x = Dense(128)(x)\n",
    "x = BatchNormalization()(x)\n",
    "x = LeakyReLU()(x)\n",
    "x = Dropout(rate = 0.5)(x)\n",
    "\n",
    "x = Dense(NUM_CLASSES)(x)\n",
    "output_layer = Activation('softmax')(x)\n",
    "\n",
    "model = Model(input_layer, output_layer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "opt = Adam(lr=0.0005)\n",
    "model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.fit(x_train\n",
    "          , y_train\n",
    "          , batch_size=32\n",
    "          , epochs=10\n",
    "          , shuffle=True\n",
    "          , validation_data = (x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.layers[6].get_weights()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.evaluate(x_test, y_test, batch_size=1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "CLASSES = np.array(['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck'])\n",
    "\n",
    "preds = model.predict(x_test)\n",
    "preds_single = CLASSES[np.argmax(preds, axis = -1)]\n",
    "actual_single = CLASSES[np.argmax(y_test, axis = -1)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "n_to_show = 10\n",
    "indices = np.random.choice(range(len(x_test)), n_to_show)\n",
    "\n",
    "fig = plt.figure(figsize=(15, 3))\n",
    "fig.subplots_adjust(hspace=0.4, wspace=0.4)\n",
    "\n",
    "for i, idx in enumerate(indices):\n",
    "    img = x_test[idx]\n",
    "    ax = fig.add_subplot(1, n_to_show, i+1)\n",
    "    ax.axis('off')\n",
    "    ax.text(0.5, -0.35, 'pred = ' + str(preds_single[idx]), fontsize=10, ha='center', transform=ax.transAxes) \n",
    "    ax.text(0.5, -0.7, 'act = ' + str(actual_single[idx]), fontsize=10, ha='center', transform=ax.transAxes)\n",
    "    ax.imshow(img)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "gdl_code",
   "language": "python",
   "name": "gdl_code"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}