Fully Connected Neural Network

H
{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Standard (Fully Connected) Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Use in Markup cell type\n",
    "#![alt text](imagename.png \"Title\")  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implementing Fully connected Neural Net"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Loading Required packages and Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "###1. Load Data and Splot Data\n",
    "from keras.datasets import mnist\n",
    "from keras.models import Sequential \n",
    "from keras.layers.core import Dense, Activation\n",
    "from keras.utils import np_utils\n",
    "(X_train, Y_train), (X_test, Y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 2000x400 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "n = 10  # how many digits we will display\n",
    "plt.figure(figsize=(20, 4))\n",
    "for i in range(n):\n",
    "    # display original\n",
    "    ax = plt.subplot(2, n, i + 1)\n",
    "    plt.imshow(X_test[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "plt.show()\n",
    "plt.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Previous X_train shape: (60000, 28, 28) \n",
      "Previous Y_train shape:(60000,)\n",
      "New X_train shape: (60000, 784) \n",
      "New Y_train shape:(60000, 10)\n"
     ]
    }
   ],
   "source": [
    "print(\"Previous X_train shape: {} \\nPrevious Y_train shape:{}\".format(X_train.shape, Y_train.shape))\n",
    "X_train = X_train.reshape(60000, 784)     \n",
    "X_test = X_test.reshape(10000, 784)\n",
    "X_train = X_train.astype('float32')     \n",
    "X_test = X_test.astype('float32')     \n",
    "X_train /= 255    \n",
    "X_test /= 255\n",
    "classes = 10\n",
    "Y_train = np_utils.to_categorical(Y_train, classes)     \n",
    "Y_test = np_utils.to_categorical(Y_test, classes)\n",
    "print(\"New X_train shape: {} \\nNew Y_train shape:{}\".format(X_train.shape, Y_train.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Setting up parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_size = 784\n",
    "batch_size = 200   \n",
    "hidden1 = 400\n",
    "hidden2 = 20\n",
    "epochs = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Building the FCN Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 400)               314000    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 20)                8020      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 10)                210       \n",
      "=================================================================\n",
      "Total params: 322,230\n",
      "Trainable params: 322,230\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "###4.Build the model\n",
    "model = Sequential()     \n",
    "model.add(Dense(hidden1, input_dim=input_size, activation='relu'))\n",
    "# output = relu (dot (W, input) + bias)\n",
    "model.add(Dense(hidden2, activation='relu'))\n",
    "model.add(Dense(classes, activation='softmax')) \n",
    "\n",
    "# Compilation\n",
    "model.compile(loss='categorical_crossentropy', \n",
    "    metrics=['accuracy'], optimizer='sgd')\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training The Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      " - 12s - loss: 1.4482 - acc: 0.6251\n",
      "Epoch 2/10\n",
      " - 3s - loss: 0.6239 - acc: 0.8482\n",
      "Epoch 3/10\n",
      " - 3s - loss: 0.4582 - acc: 0.8798\n",
      "Epoch 4/10\n",
      " - 3s - loss: 0.3941 - acc: 0.8936\n",
      "Epoch 5/10\n",
      " - 3s - loss: 0.3579 - acc: 0.9011\n",
      "Epoch 6/10\n",
      " - 4s - loss: 0.3328 - acc: 0.9070\n",
      "Epoch 7/10\n",
      " - 3s - loss: 0.3138 - acc: 0.9118\n",
      "Epoch 8/10\n",
      " - 3s - loss: 0.2980 - acc: 0.9157\n",
      "Epoch 9/10\n",
      " - 3s - loss: 0.2849 - acc: 0.9191\n",
      "Epoch 10/10\n",
      " - 3s - loss: 0.2733 - acc: 0.9223\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x272375a7240>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fitting on Data\n",
    "model.fit(X_train, Y_train, batch_size=batch_size, epochs=10, verbose=2)\n",
    "###5.Test "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "#### Testing The Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 121us/step\n",
      "\n",
      "Test accuracy: 0.9257\n",
      "[0 6 9 0 1 5 9 7 3 4]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "score = model.evaluate(X_test, Y_test, verbose=1)\n",
    "print('\\n''Test accuracy:', score[1])\n",
    "mask = range(10,20)\n",
    "X_valid = X_test[mask]\n",
    "y_pred = model.predict_classes(X_valid)\n",
    "print(y_pred)\n",
    "plt.figure(figsize=(20, 4))\n",
    "for i in range(n):\n",
    "    # display original\n",
    "    ax = plt.subplot(2, n, i + 1)\n",
    "    plt.imshow(X_valid[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "plt.show()\n",
    "plt.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}
About this Algorithm

Standard (Fully Connected) Neural Network

#Use in Markup cell type
#![alt text](imagename.png "Title")  

Implementing Fully connected Neural Net

Loading Required packages and Data

###1. Load Data and Splot Data
from keras.datasets import mnist
from keras.models import Sequential 
from keras.layers.core import Dense, Activation
from keras.utils import np_utils
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()
Using TensorFlow backend.

Preprocessing

import matplotlib.pyplot as plt
n = 10  # how many digits we will display
plt.figure(figsize=(20, 4))
for i in range(n):
    # display original
    ax = plt.subplot(2, n, i + 1)
    plt.imshow(X_test[i].reshape(28, 28))
    plt.gray()
    ax.get_xaxis().set_visible(False)
    ax.get_yaxis().set_visible(False)
plt.show()
plt.close()
&lt;Figure size 2000x400 with 10 Axes&gt;
print("Previous X_train shape: {} \nPrevious Y_train shape:{}".format(X_train.shape, Y_train.shape))
X_train = X_train.reshape(60000, 784)     
X_test = X_test.reshape(10000, 784)
X_train = X_train.astype('float32')     
X_test = X_test.astype('float32')     
X_train /= 255    
X_test /= 255
classes = 10
Y_train = np_utils.to_categorical(Y_train, classes)     
Y_test = np_utils.to_categorical(Y_test, classes)
print("New X_train shape: {} \nNew Y_train shape:{}".format(X_train.shape, Y_train.shape))
Previous X_train shape: (60000, 28, 28) 
Previous Y_train shape:(60000,)
New X_train shape: (60000, 784) 
New Y_train shape:(60000, 10)

Setting up parameters

input_size = 784
batch_size = 200   
hidden1 = 400
hidden2 = 20
epochs = 2

Building the FCN Model

###4.Build the model
model = Sequential()     
model.add(Dense(hidden1, input_dim=input_size, activation='relu'))
# output = relu (dot (W, input) + bias)
model.add(Dense(hidden2, activation='relu'))
model.add(Dense(classes, activation='softmax')) 

# Compilation
model.compile(loss='categorical_crossentropy', 
    metrics=['accuracy'], optimizer='sgd')
model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_1 (Dense)              (None, 400)               314000    
_________________________________________________________________
dense_2 (Dense)              (None, 20)                8020      
_________________________________________________________________
dense_3 (Dense)              (None, 10)                210       
=================================================================
Total params: 322,230
Trainable params: 322,230
Non-trainable params: 0
_________________________________________________________________

Training The Model

# Fitting on Data
model.fit(X_train, Y_train, batch_size=batch_size, epochs=10, verbose=2)
###5.Test 
Epoch 1/10
 - 12s - loss: 1.4482 - acc: 0.6251
Epoch 2/10
 - 3s - loss: 0.6239 - acc: 0.8482
Epoch 3/10
 - 3s - loss: 0.4582 - acc: 0.8798
Epoch 4/10
 - 3s - loss: 0.3941 - acc: 0.8936
Epoch 5/10
 - 3s - loss: 0.3579 - acc: 0.9011
Epoch 6/10
 - 4s - loss: 0.3328 - acc: 0.9070
Epoch 7/10
 - 3s - loss: 0.3138 - acc: 0.9118
Epoch 8/10
 - 3s - loss: 0.2980 - acc: 0.9157
Epoch 9/10
 - 3s - loss: 0.2849 - acc: 0.9191
Epoch 10/10
 - 3s - loss: 0.2733 - acc: 0.9223
&lt;keras.callbacks.History at 0x272375a7240&gt;

Testing The Model

score = model.evaluate(X_test, Y_test, verbose=1)
print('\n''Test accuracy:', score[1])
mask = range(10,20)
X_valid = X_test[mask]
y_pred = model.predict_classes(X_valid)
print(y_pred)
plt.figure(figsize=(20, 4))
for i in range(n):
    # display original
    ax = plt.subplot(2, n, i + 1)
    plt.imshow(X_valid[i].reshape(28, 28))
    plt.gray()
    ax.get_xaxis().set_visible(False)
    ax.get_yaxis().set_visible(False)
plt.show()
plt.close()
10000/10000 [==============================] - 1s 121us/step

Test accuracy: 0.9257
[0 6 9 0 1 5 9 7 3 4]