mnist数据集实践（tensorflow实现）

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

MNIST_data_folder="E:\Code\jupyter_notebook_file\Tensorflow\MNIST_data"
mnist = input_data.read_data_sets(MNIST_data_folder, one_hot=True)

batch_size = 100
# 计算一共多少个批次
n_batch = mnist.train.num_examples // batch_size

# 初始化权值
def init_weight_varible(shape):
    initial = tf.truncated_normal(shape, stddev=0.1)
    return tf.Variable(initial)
# 初始化偏置
def init_bias_variable(shape):
    initial = tf.constant(0.1, shape=shape)
    return tf.Variable(initial)
# 卷积层
def con2d(x, W):
    return tf.nn.conv2d(x, W, strides=[1,1,1,1], padding='SAME')
# 池化层
def max_pool_2x2(x):
    return tf.nn.max_pool(x, ksize=[1,2,2,1], strides=[1,2,2,1],padding='SAME')

# 定义两个占位
x = tf.placeholder(tf.float32, shape=(None, 784))
y = tf.placeholder(tf.float32, shape=(None, 10))

#转化为4D
x_image= tf.reshape(x, [-1, 28, 28, 1])
# 初始化第一个卷积层的权值和偏置
W_conv1 = init_weight_varible([5,5,1,32])
b_conv1 = init_bias_variable([32])
# 将x_image进行卷积和加上偏置 在用激活函数
h_conv1 = tf.nn.relu(con2d(x_image, W_conv1) + b_conv1)      #28，28，32
h_pool1 = max_pool_2x2(h_conv1)#池化 #14，14, 32 

#第二个卷积层
W_conv2 = init_weight_varible([5,5,32,64])                   
b_conv2 = init_bias_variable([64])

h_conv2 = tf.nn.relu(con2d(h_pool1, W_conv2 )+ b_conv2)      #14 14 64
h_pool2 = max_pool_2x2(h_conv2)                              #7 7 64

h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])

W_fc1 = init_weight_varible([7*7*64, 1024])
b_fcl = init_bias_variable([1024])


h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fcl)

keep_drop = tf.placeholder(tf.float32)
h_fcl_drop = tf.nn.dropout(h_fc1, keep_drop)

W_fc2 = init_weight_varible([1024, 10])
b_fc2 = init_bias_variable([10])

prediction = tf.nn.softmax(tf.matmul(h_fcl_drop, W_fc2) + b_fc2)

cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y, logits=prediction))

train_step = tf.train.GradientDescentOptimizer(0.1).minimize(cross_entropy)

correct_pre = tf.equal(tf.argmax(prediction, 1), tf.argmax(y, 1))#返回最大值的索引， axis=1 按行取值

accuracy = tf.reduce_mean(tf.cast(correct_pre, tf.float32))

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    for epoch in range(21):
# 控制学习率 assign 赋值
        for batch in range(n_batch):
            batch_xs, batch_ys = mnist.train.next_batch(batch_size)
            sess.run(train_step, feed_dict={x:batch_xs, y:batch_ys, keep_drop:0.7})
        TestAcc = sess.run(accuracy, feed_dict={x:mnist.test.images, y:mnist.test.labels, keep_drop: 1})
        
        print('Iter' + str(epoch) + ', test acc' + str(TestAcc) + ', train acc' + str(TrainAcc))