TensorFlow tutorial: Small CNN to process MNIST data

mnist_cnn.py

# From instructions at https://www.tensorflow.org/versions/r1.0/get_started/mnist/pros

import argparse
import sys
import tensorflow as tf
import time

from tensorflow.examples.tutorials.mnist import input_data
FLAGS = None


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)


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')


def main(_):

    mnist = input_data.read_data_sets(FLAGS.data_dir)

    # ---------------- MODEL DEF -------------------
    x = tf.placeholder(tf.float32, [None, 784])
    y_ = tf.placeholder(tf.int64, [None])

    W = tf.Variable(tf.zeros([784, 10]))
    b = tf.Variable(tf.zeros([10]))

    # --- 1 --- define our first convolutional layer
    W_conv1 = weight_variable([5, 5, 1, 32])
    b_conv1 = bias_variable([32])
    x_image = tf.reshape(x, [-1, 28, 28, 1])

    h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
    h_pool1 = max_pool_2x2(h_conv1)
    # image is now 14x14

    # --- 2 --- DEFINE 2nd convolutional layer
    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)
    # image is now 7x7

    # --- 3 --- DENSELY CONNECTED LAYER
    # fully connected, 1024 neurons

    W_fc1 = weight_variable([7*7*64, 1024])
    # input is 7x7 x 64 channels
    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)

    # --- 4 --- DROPOUT
    keep_prob = tf.placeholder(tf.float32)
    h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)

    # --- 5 --- READOUT LAYER: readout to our 10 values in the one-hot label

    W_fc2 = weight_variable([1024, 10])
    b_fc2 = bias_variable([10])

    y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2


    # --------------- TRAINING STEPS -----------------

    cross_entropy = tf.reduce_mean(tf.losses.sparse_softmax_cross_entropy(labels=y_, logits=y_conv))
    train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)

    # evaluating our model
    correct_prediction = tf.equal(tf.argmax(y_conv, 1), y_)
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

    # now run it!
    sess = tf.InteractiveSession()
    sess.run(tf.global_variables_initializer())

    # initialize all variables
    tf.global_variables_initializer().run()

    # run training step many times!
    start = time.time()
    for i in range(1000):
        batch = mnist.train.next_batch(50)
        if i%100 == 0:
            train_accuracy = accuracy.eval(feed_dict={x:batch[0], y_:batch[1], keep_prob: 1.0})
            print("Step %d, training accuracy %g"%(i, train_accuracy))
        train_step.run(feed_dict={x:batch[0], y_:batch[1], keep_prob: 0.5})
    end = time.time()

    # print how well the model does on the test data
    print("test accuracy %g"%accuracy.eval(feed_dict={
        x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0
    }))
    print("Training time: {} sec".format(end - start))


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument(
        '--data_dir',
        type=str,
        default='/tmp/tensorflow/mnist/input_data',
        help='Directory for storing input data')
    FLAGS, unparsed = parser.parse_known_args()
    tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)