# SINGLE LAYER AND MULTI LAYER NETWORKS FOR MNIST
# BASED ON CODE FROM TENSORFLOW TUTORIAL
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
# MODEL
# CREATE PLACEHOLDER VARIABLES FOR OPERATION MANIPULATION
# THE 784 MATCHES THE VECTOR SIZE OF THE MNIST IMAGES - 28*28
x = tf.placeholder(tf.float32, [None, 784])
# MODEL
# CREATE WEIGHTS & BIASES VARIABLES
# IN TF, OUR MODEL PARAMETERS ARE OFTEN MANAGED AS VARIABLES
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
# MODEL
# CREATE MODEL - USES SOFTMAX AS THE ACTIVATION FUNCTION
# REMEMBER GOAL FOR ACTIVATION FUNCTION IS TO "SHAPE" THE
# OUTPUT INTO A PROBABILITY DISTRO OVER THE 10 CLASSES
y = tf.nn.softmax(tf.matmul(x, W) + b)
# MODEL
# CREATE PREDICTED VARIABLE Y-HAT
# AND USE CROSS ENTROPY TO DETERMINE LOSS
# CROSS ENTROPY - HOW INEFFICIENT ARE OUR PREDICTIONS?
y_ = tf.placeholder(tf.float32, [None, 10])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))
# MODEL
# TRAIN USING GRADIENT DESCENT
# LEARNING RATE AT MIDPOINT - .5 - MAKE SMALL ADJUSTMENTS TO MINIMIZE COST
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
# MODEL - RUN
sess = tf.InteractiveSession()
tf.global_variables_initializer().run()
for _ in range(10000):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})
# EVALUATE MODEL
correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels}))
Alternative Approach
# WEIGHT INITIALIZATION
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
# CREATE CONVOLUTION AND POOLING LAYERS
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
strides=[1, 2, 2, 1], padding='SAME')
# FIRST CONVOLUTION LAYER
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
x_image = tf.reshape(x, [-1, 28, 28, 1]) # BASE IMAGE SIZE OF 28 * 28
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1) # RESULTING IMAGE SIZE IS 14 * 14
# SECOND CONOLUTION LAYER
# MORE THAN ONE LAYER? DEEP LEARNING
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
# FULLY CONNECTED LAYER - BEFORE OUTPUT
W_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1) # ADD THE RECTIFIED LINEAR UNIT
# DROP LAYER - REDUCE OVERFITTING
# USE OF rate IS AN UPDATE BASED ON TF
keep_prob = tf.placeholder(tf.float32)
rate = 1 - keep_prob
h_fc1_drop = tf.nn.dropout(h_fc1, rate)
# LAST LAYER - OUTPUT
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
# RUN THE MODEL
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
for i in range(10000):
batch = mnist.train.next_batch(50)
if i % 100 == 0:
train_accuracy = accuracy.eval(feed_dict={
x: batch[0], y_: batch[1], rate: 1.0})
print('step %d, training accuracy %g' % (i, train_accuracy))
train_step.run(feed_dict={x: batch[0], y_: batch[1], rate: 0.5})
print('test accuracy %g' % accuracy.eval(feed_dict={
x: mnist.test.images, y_: mnist.test.labels, rate: 1.0}))