TensorFlow--MNIST

1 Requirement

• Python 3.5
• TensorFlow 1.4.0

2 Source code

2.1 layers.py

import tensorflow as tf


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)

# strides=[batch, height, width, channels]
def conv2d(x, W):
    return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')

# ksize=[batch, height, width, channels]
def max_pool_2x2(x):
    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
                          strides=[1, 2, 2, 1], padding='SAME')


def conv_layer(input, shape):
    W = weight_variable(shape)
    b = bias_variable([shape[3]])
    return tf.nn.relu(conv2d(input, W) + b)


def full_layer(input, size):
    in_size = int(input.get_shape()[1])
    W = weight_variable([in_size, size])
    b = bias_variable([size])
    return tf.matmul(input, W) + b

2.2 mnist_cnn.py

from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf
import numpy as np

from layers import conv_layer, max_pool_2x2, full_layer

DATA_DIR = '/tmp/data'
MINIBATCH_SIZE = 50
STEPS = 5000


mnist = input_data.read_data_sets(DATA_DIR, one_hot=True)

x = tf.placeholder(tf.float32, shape=[None, 784])
y_ = tf.placeholder(tf.float32, shape=[None, 10])

# -1 auto
x_image = tf.reshape(x, [-1, 28, 28, 1])
# 5x5 1 channel 32 feature maps
conv1 = conv_layer(x_image, shape=[5, 5, 1, 32])
conv1_pool = max_pool_2x2(conv1)

# 5x5 32 channels 64 feature maps
conv2 = conv_layer(conv1_pool, shape=[5, 5, 32, 64])
conv2_pool = max_pool_2x2(conv2)

# -1 auto
# 28 -> 14 -> 7 
conv2_flat = tf.reshape(conv2_pool, [-1, 7*7*64])
full_1 = tf.nn.relu(full_layer(conv2_flat, 1024))

keep_prob = tf.placeholder(tf.float32)
full1_drop = tf.nn.dropout(full_1, keep_prob=keep_prob)

y_conv = full_layer(full1_drop, 10)

cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=y_conv, labels=y_))
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(STEPS):
        batch = mnist.train.next_batch(MINIBATCH_SIZE)

        if i % 100 == 0:
            train_accuracy = sess.run(accuracy, feed_dict={x: batch[0], y_: batch[1],
                                                           keep_prob: 1.0})
            print("step {}, training accuracy {}".format(i, train_accuracy))

        sess.run(train_step, feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})
    # 10 blocks of 1000 images each. 784(28*28)
    X = mnist.test.images.reshape(10, 1000, 784)
    Y = mnist.test.labels.reshape(10, 1000, 10)
    test_accuracy = np.mean(
        [sess.run(accuracy, feed_dict={x: X[i], y_: Y[i], keep_prob: 1.0}) for i in range(10)])

print("test accuracy: {}".format(test_accuracy))

3 Result

Extracting /tmp/data/train-images-idx3-ubyte.gz
Extracting /tmp/data/train-labels-idx1-ubyte.gz
Extracting /tmp/data/t10k-images-idx3-ubyte.gz
Extracting /tmp/data/t10k-labels-idx1-ubyte.gz
2018-07-10 17:19:02.125927: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
2018-07-10 17:19:02.197368: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:892] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-07-10 17:19:02.197689: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Found device 0 with properties:
name: GeForce GTX TITAN X major: 5 minor: 2 memoryClockRate(GHz): 1.076
pciBusID: 0000:01:00.0
totalMemory: 11.92GiB freeMemory: 11.66GiB
2018-07-10 17:19:02.197705: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX TITAN X, pci bus id: 0000:01:00.0, compute capability: 5.2)
step 0, training accuracy 0.219999998808
step 100, training accuracy 0.699999988079
step 200, training accuracy 0.759999990463
step 300, training accuracy 0.920000016689
step 400, training accuracy 0.879999995232
step 500, training accuracy 0.939999997616
step 600, training accuracy 0.939999997616
step 700, training accuracy 0.980000019073
step 800, training accuracy 0.980000019073
step 900, training accuracy 0.939999997616
step 1000, training accuracy 0.939999997616
step 1100, training accuracy 0.939999997616
step 1200, training accuracy 0.980000019073
step 1300, training accuracy 0.959999978542
step 1400, training accuracy 0.959999978542
step 1500, training accuracy 0.959999978542
step 1600, training accuracy 0.959999978542
step 1700, training accuracy 0.939999997616
step 1800, training accuracy 0.980000019073
step 1900, training accuracy 1.0
step 2000, training accuracy 1.0
step 2100, training accuracy 1.0
step 2200, training accuracy 0.980000019073
step 2300, training accuracy 0.939999997616
step 2400, training accuracy 0.980000019073
step 2500, training accuracy 0.980000019073
step 2600, training accuracy 1.0
step 2700, training accuracy 0.980000019073
step 2800, training accuracy 0.959999978542
step 2900, training accuracy 0.980000019073
step 3000, training accuracy 0.959999978542
step 3100, training accuracy 1.0
step 3200, training accuracy 0.980000019073
step 3300, training accuracy 1.0
step 3400, training accuracy 1.0
step 3500, training accuracy 0.959999978542
step 3600, training accuracy 0.980000019073
step 3700, training accuracy 1.0
step 3800, training accuracy 0.959999978542
step 3900, training accuracy 0.980000019073
step 4000, training accuracy 0.959999978542
step 4100, training accuracy 0.980000019073
step 4200, training accuracy 0.980000019073
step 4300, training accuracy 1.0
step 4400, training accuracy 0.980000019073
step 4500, training accuracy 0.939999997616
step 4600, training accuracy 1.0
step 4700, training accuracy 0.980000019073
step 4800, training accuracy 0.959999978542
step 4900, training accuracy 1.0
test accuracy: 0.986999988556

4 Reference

• Oreilly-Learning-TensorFlow

• Learning TensorFlow

• TensorFlow strides 参数讨论

• 【TensorFlow】tf.nn.max_pool实现池化操作