1.+Cifar_10_with_dropout_without_BN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# import libraries\n",
    "from __future__ import print_function\n",
    "import keras\n",
    "from keras.datasets import cifar10\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "import numpy as np\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# batch, classes, epochs\n",
    "batch_size = 32\n",
    "num_classes = 10\n",
    "epochs = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
      "170500096/170498071 [==============================] - 12s 0us/step\n",
      "x_train shape: (50000, 32, 32, 3)\n",
      "50000 train samples\n",
      "10000 test samples\n"
     ]
    }
   ],
   "source": [
    "# The data, split between train and test sets:\n",
    "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
    "print('x_train shape:', x_train.shape)\n",
    "print(x_train.shape[0], 'train samples')\n",
    "print(x_test.shape[0], 'test samples')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x216 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plotting some random 10 images\n",
    "class_names = ['airplane','automobile','bird','cat','deer',\n",
    "               'dog','frog','horse','ship','truck']\n",
    "\n",
    "fig = plt.figure(figsize=(8,3))\n",
    "for i in range(num_classes):\n",
    "    ax = fig.add_subplot(2, 5, 1 + i, xticks=[], yticks=[])\n",
    "    idx = np.where(y_train[:]==i)[0]\n",
    "    features_idx = x_train[idx,::]\n",
    "    img_num = np.random.randint(features_idx.shape[0])\n",
    "    im = (features_idx[img_num,::])\n",
    "    ax.set_title(class_names[i])\n",
    "    plt.imshow(im)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert class vectors to binary class matrices.\n",
    "y_train = keras.utils.to_categorical(y_train, num_classes)\n",
    "y_test = keras.utils.to_categorical(y_test, num_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model architecture\n",
    "model = Sequential()\n",
    "model.add(Conv2D(32, (3, 3), padding='same',\n",
    "                 input_shape=x_train.shape[1:]))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(32, (3, 3)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "model.add(Dropout(0.25))\n",
    "\n",
    "model.add(Conv2D(64, (3, 3), padding='same'))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(64, (3, 3)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "model.add(Dropout(0.25))\n",
    "\n",
    "model.add(Flatten())\n",
    "model.add(Dense(512))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(num_classes))\n",
    "model.add(Activation('softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 32, 32, 32)        896       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 32, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 30, 30, 32)        9248      \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 30, 30, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 15, 15, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 15, 15, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 15, 15, 64)        18496     \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 15, 15, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 13, 13, 64)        36928     \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 13, 13, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 6, 6, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 6, 6, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 2304)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 512)               1180160   \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                5130      \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 10)                0         \n",
      "=================================================================\n",
      "Total params: 1,250,858\n",
      "Trainable params: 1,250,858\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# summary\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# compile the model\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='sgd',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "# convert to float, normalise the data\n",
    "x_train = x_train.astype('float32')\n",
    "x_test = x_test.astype('float32')\n",
    "x_train /= 255\n",
    "x_test /= 255\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 50000 samples, validate on 10000 samples\n",
      "Epoch 1/50\n",
      "50000/50000 [==============================] - 47s 942us/step - loss: 2.0599 - acc: 0.2338 - val_loss: 1.7676 - val_acc: 0.3740\n",
      "Epoch 2/50\n",
      "50000/50000 [==============================] - 19s 389us/step - loss: 1.7274 - acc: 0.3714 - val_loss: 1.5703 - val_acc: 0.4314\n",
      "Epoch 3/50\n",
      "50000/50000 [==============================] - 19s 389us/step - loss: 1.5755 - acc: 0.4274 - val_loss: 1.5418 - val_acc: 0.4422\n",
      "Epoch 4/50\n",
      "50000/50000 [==============================] - 19s 379us/step - loss: 1.4547 - acc: 0.4730 - val_loss: 1.3025 - val_acc: 0.5288\n",
      "Epoch 5/50\n",
      "50000/50000 [==============================] - 18s 370us/step - loss: 1.3665 - acc: 0.5095 - val_loss: 1.2580 - val_acc: 0.5595\n",
      "Epoch 6/50\n",
      "50000/50000 [==============================] - 19s 378us/step - loss: 1.2981 - acc: 0.5355 - val_loss: 1.1751 - val_acc: 0.5851\n",
      "Epoch 7/50\n",
      "50000/50000 [==============================] - 19s 371us/step - loss: 1.2343 - acc: 0.5617 - val_loss: 1.1319 - val_acc: 0.6062\n",
      "Epoch 8/50\n",
      "50000/50000 [==============================] - 19s 373us/step - loss: 1.1793 - acc: 0.5817 - val_loss: 1.0903 - val_acc: 0.6124\n",
      "Epoch 9/50\n",
      "50000/50000 [==============================] - 19s 371us/step - loss: 1.1316 - acc: 0.5967 - val_loss: 1.0624 - val_acc: 0.6270\n",
      "Epoch 10/50\n",
      "50000/50000 [==============================] - 19s 373us/step - loss: 1.0921 - acc: 0.6159 - val_loss: 1.0050 - val_acc: 0.6495\n",
      "Epoch 11/50\n",
      "50000/50000 [==============================] - 19s 371us/step - loss: 1.0522 - acc: 0.6288 - val_loss: 0.9769 - val_acc: 0.6555\n",
      "Epoch 12/50\n",
      "50000/50000 [==============================] - 19s 371us/step - loss: 1.0152 - acc: 0.6400 - val_loss: 0.9389 - val_acc: 0.6781\n",
      "Epoch 13/50\n",
      "50000/50000 [==============================] - 19s 372us/step - loss: 0.9812 - acc: 0.6571 - val_loss: 0.9574 - val_acc: 0.6655\n",
      "Epoch 14/50\n",
      "50000/50000 [==============================] - 18s 370us/step - loss: 0.9530 - acc: 0.6619 - val_loss: 0.9401 - val_acc: 0.6724\n",
      "Epoch 15/50\n",
      "50000/50000 [==============================] - 18s 365us/step - loss: 0.9211 - acc: 0.6756 - val_loss: 0.8555 - val_acc: 0.7006\n",
      "Epoch 16/50\n",
      "50000/50000 [==============================] - 18s 352us/step - loss: 0.8967 - acc: 0.6821 - val_loss: 0.8683 - val_acc: 0.6964\n",
      "Epoch 17/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.8685 - acc: 0.6929 - val_loss: 0.8191 - val_acc: 0.7145\n",
      "Epoch 18/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.8466 - acc: 0.7007 - val_loss: 0.8218 - val_acc: 0.7137\n",
      "Epoch 19/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.8275 - acc: 0.7070 - val_loss: 0.7869 - val_acc: 0.7266\n",
      "Epoch 20/50\n",
      "50000/50000 [==============================] - 17s 348us/step - loss: 0.8041 - acc: 0.7161 - val_loss: 0.7856 - val_acc: 0.7303\n",
      "Epoch 21/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.7856 - acc: 0.7223 - val_loss: 0.7480 - val_acc: 0.7398\n",
      "Epoch 22/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.7638 - acc: 0.7303 - val_loss: 0.7574 - val_acc: 0.7397\n",
      "Epoch 23/50\n",
      "50000/50000 [==============================] - 18s 354us/step - loss: 0.7446 - acc: 0.7350 - val_loss: 0.7312 - val_acc: 0.7471\n",
      "Epoch 24/50\n",
      "50000/50000 [==============================] - 17s 348us/step - loss: 0.7223 - acc: 0.7459 - val_loss: 0.7057 - val_acc: 0.7590\n",
      "Epoch 25/50\n",
      "50000/50000 [==============================] - 17s 350us/step - loss: 0.7095 - acc: 0.7497 - val_loss: 0.7037 - val_acc: 0.7555\n",
      "Epoch 26/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.6910 - acc: 0.7549 - val_loss: 0.6918 - val_acc: 0.7633\n",
      "Epoch 27/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.6810 - acc: 0.7583 - val_loss: 0.7839 - val_acc: 0.7292\n",
      "Epoch 28/50\n",
      "50000/50000 [==============================] - 17s 349us/step - loss: 0.6634 - acc: 0.7668 - val_loss: 0.7116 - val_acc: 0.7607\n",
      "Epoch 29/50\n",
      "50000/50000 [==============================] - 17s 345us/step - loss: 0.6528 - acc: 0.7706 - val_loss: 0.6817 - val_acc: 0.7659\n",
      "Epoch 30/50\n",
      "50000/50000 [==============================] - 17s 348us/step - loss: 0.6337 - acc: 0.7781 - val_loss: 0.6622 - val_acc: 0.7676\n",
      "Epoch 31/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.6222 - acc: 0.7798 - val_loss: 0.6520 - val_acc: 0.7779\n",
      "Epoch 32/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.6102 - acc: 0.7821 - val_loss: 0.6467 - val_acc: 0.7766\n",
      "Epoch 33/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.5913 - acc: 0.7916 - val_loss: 0.6416 - val_acc: 0.7790\n",
      "Epoch 34/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.5816 - acc: 0.7950 - val_loss: 0.6374 - val_acc: 0.7796\n",
      "Epoch 35/50\n",
      "50000/50000 [==============================] - 17s 345us/step - loss: 0.5683 - acc: 0.7979 - val_loss: 0.6228 - val_acc: 0.7861\n",
      "Epoch 36/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.5631 - acc: 0.7991 - val_loss: 0.6328 - val_acc: 0.7825\n",
      "Epoch 37/50\n",
      "50000/50000 [==============================] - 17s 348us/step - loss: 0.5489 - acc: 0.8039 - val_loss: 0.6239 - val_acc: 0.7865\n",
      "Epoch 38/50\n",
      "50000/50000 [==============================] - 17s 344us/step - loss: 0.5338 - acc: 0.8097 - val_loss: 0.6221 - val_acc: 0.7832\n",
      "Epoch 39/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.5298 - acc: 0.8123 - val_loss: 0.6142 - val_acc: 0.7918\n",
      "Epoch 40/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.5185 - acc: 0.8157 - val_loss: 0.6033 - val_acc: 0.7926\n",
      "Epoch 41/50\n",
      "50000/50000 [==============================] - 18s 355us/step - loss: 0.5055 - acc: 0.8193 - val_loss: 0.6108 - val_acc: 0.7881\n",
      "Epoch 42/50\n",
      "50000/50000 [==============================] - 17s 349us/step - loss: 0.4981 - acc: 0.8212 - val_loss: 0.5996 - val_acc: 0.7934\n",
      "Epoch 43/50\n",
      "50000/50000 [==============================] - 17s 349us/step - loss: 0.4889 - acc: 0.8249 - val_loss: 0.6029 - val_acc: 0.7930\n",
      "Epoch 44/50\n",
      "50000/50000 [==============================] - 17s 347us/step - loss: 0.4794 - acc: 0.8282 - val_loss: 0.6147 - val_acc: 0.7959\n",
      "Epoch 45/50\n",
      "50000/50000 [==============================] - 17s 344us/step - loss: 0.4692 - acc: 0.8332 - val_loss: 0.5915 - val_acc: 0.7982\n",
      "Epoch 46/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.4611 - acc: 0.8351 - val_loss: 0.6250 - val_acc: 0.7903\n",
      "Epoch 47/50\n",
      "50000/50000 [==============================] - 17s 348us/step - loss: 0.4520 - acc: 0.8380 - val_loss: 0.6274 - val_acc: 0.7899\n",
      "Epoch 48/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.4468 - acc: 0.8413 - val_loss: 0.6091 - val_acc: 0.7919\n",
      "Epoch 49/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.4352 - acc: 0.8448 - val_loss: 0.6070 - val_acc: 0.7979\n",
      "Epoch 50/50\n",
      "50000/50000 [==============================] - 17s 346us/step - loss: 0.4273 - acc: 0.8459 - val_loss: 0.5923 - val_acc: 0.7982\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f68d198b9e8>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# train \n",
    "model.fit(x_train, y_train,\n",
    "              batch_size=batch_size,\n",
    "              epochs=epochs,\n",
    "              validation_data=(x_test, y_test),\n",
    "              shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Result: After 50 epochs, the difference between training and validation accuracy is low but the training is slow as compared to the model with batch normalization."
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}