MNIST with CNN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Import the necessary libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from keras.datasets import mnist\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
    "from keras.optimizers import Adam\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.utils import np_utils\n",
    "from keras.layers import Conv2D, MaxPooling2D, ZeroPadding2D, GlobalAveragePooling2D\n",
    "from keras.layers.advanced_activations import LeakyReLU \n",
    "from keras.preprocessing.image import ImageDataGenerator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Load the MNIST data provided by keras\n",
    "(X_train, y_train), (X_test, y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# The shape of each array is (60000, 28, 28)\n",
    "# Since we are using tensorflow the format of each array should be (batch, height, width, channels)\n",
    "# Let's reshape it then!\n",
    "X_train = X_train.reshape(X_train.shape[0], 28, 28, 1)\n",
    "X_test = X_test.reshape(X_test.shape[0], 28, 28, 1)\n",
    "\n",
    "X_train = X_train.astype('float32')\n",
    "X_test = X_test.astype('float32')\n",
    "\n",
    "X_train/=255\n",
    "X_test/=255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Apply one-hot encoding\n",
    "number_of_classes = 10\n",
    "\n",
    "Y_train = np_utils.to_categorical(y_train, number_of_classes)\n",
    "Y_test = np_utils.to_categorical(y_test, number_of_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Now let's create the CNN model that will classify the MNIST images\n",
    "\n",
    "# Steps:\n",
    "# 1. Convolution\n",
    "# 2. Activation\n",
    "# 3. Pooling\n",
    "# 4. Repetition of the previous steps to add more hidden layers\n",
    "# 5. Create a fully connected network\n",
    "\n",
    "model = Sequential()\n",
    "\n",
    "model.add(Conv2D(32, (3, 3), input_shape=(28,28,1)))\n",
    "BatchNormalization(axis=-1)\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(32, (3, 3)))\n",
    "BatchNormalization(axis=-1)\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(64,(3, 3)))\n",
    "BatchNormalization(axis=-1)\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(64, (3, 3)))\n",
    "BatchNormalization(axis=-1)\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())\n",
    "\n",
    "# Fully connected layer\n",
    "model.add(Dense(512))\n",
    "BatchNormalization()\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(10))\n",
    "\n",
    "model.add(Activation('softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Compile the model\n",
    "model.compile(loss='categorical_crossentropy', optimizer=Adam(), metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Apply data augmentation to reduce over-fitting\n",
    "gen = ImageDataGenerator(rotation_range=8, width_shift_range=0.08, shear_range=0.3,\n",
    "                         height_shift_range=0.08, zoom_range=0.08)\n",
    "\n",
    "test_gen = ImageDataGenerator()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Create batches in order to use less memory.\n",
    "# Using batch of 64, the model will take 64 images at a time in the process of training\n",
    "train_generator = gen.flow(X_train, Y_train, batch_size=64)\n",
    "test_generator = test_gen.flow(X_test, Y_test, batch_size=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "937/937 [==============================] - 290s - loss: 0.2359 - acc: 0.9242 - val_loss: 0.0331 - val_acc: 0.9910\n",
      "Epoch 2/5\n",
      "937/937 [==============================] - 292s - loss: 0.0776 - acc: 0.9757 - val_loss: 0.0280 - val_acc: 0.9905\n",
      "Epoch 3/5\n",
      "937/937 [==============================] - 297s - loss: 0.0591 - acc: 0.9817 - val_loss: 0.0197 - val_acc: 0.9935\n",
      "Epoch 4/5\n",
      "937/937 [==============================] - 274s - loss: 0.0507 - acc: 0.9843 - val_loss: 0.0258 - val_acc: 0.9923\n",
      "Epoch 5/5\n",
      "937/937 [==============================] - 276s - loss: 0.0452 - acc: 0.9859 - val_loss: 0.0182 - val_acc: 0.9937\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1a2ebbbe48>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We're ready to train the model\n",
    "# shuffle argument can be used here as well\n",
    "model.fit_generator(train_generator, steps_per_epoch=60000//64, epochs=5, \n",
    "                    validation_data=test_generator, validation_steps=10000//64)"
   ]
  }
 ],
 "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": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}