tensorflow学习（四）设置两层神经网络对mnist数据进行十分类

import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials import input_data
mnist = input_data.read_data_sets('data/', one_hot= True)


#定义神经网络各层的神经元数量
hidden1 = 256
hidden2 = 128
num_input = 784
num_output = 10

#设置输出和输入
x = tf.placeholder('float' , [None,num_input])
y = tf.placeholder('float', [None, num_output])
#初始化权重
std = 0.1
w = {
    'w1': tf.Variable(tf.random.normal([num_input, hidden1],stddev=std)),
    'w2': tf.Variable(tf.random.normal([hidden1, hidden2],stddev=std)),
    'w3': tf.Variable(tf.random.normal([hidden2, num_output],stddev=std))
}



b = {
    'b1': tf.Variable(tf.random.normal([hidden1],stddev=std)),
    'b2': tf.Variable(tf.random.normal([hidden2],stddev=std)),
    'b3': tf.Variable(tf.random.normal([num_output],stddev=std))
   }

#定义网络

def  Lnet(X, weight, biases):
    layer1 = tf.nn.sigmoid(tf.add(tf.matmul(X,weight['w1']),biases['b1']))
    layer2 = tf.nn.sigmoid(tf.add(tf.matmul(layer1, weight['w2']), biases['b2']))
    out1 = tf.matmul(layer2, weight['w3'])+biases['b3']
    return out1


#定义反向误差传播

pre = Lnet(x, w, b)

loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels= y, logits=pre))
optimizer = tf.train.GradientDescentOptimizer(0.001).minimize(loss)
correct = tf.equal(tf.argmax(pre, 1), tf.argmax(y, 1))
accr = tf.reduce_mean(tf.cast(correct,'float'))

init = tf.global_variables_initializer()

batch_size =100

with tf.Session() as sess:
    sess.run(init)
    for epoch in range(20):
        avg_cost = 0
        num_batch = int(mnist.train.num_examples/batch_size)
        for i in range(num_batch):
            batch_xs, batch_ys = mnist.train.next_batch(batch_size)

            sess.run(optimizer, feed_dict={x:batch_xs, y:batch_ys})

            avg_cost += sess.run(loss, feed_dict={x:batch_xs, y:batch_ys})
        avg_cost = avg_cost/num_batch

        if (epoch+1) %5 ==0:

            feeds_train = {x:batch_xs, y:batch_ys}

            feeds_test = {x:mnist.test.images, y: mnist.test.labels}
            train_acc = sess.run(accr, feed_dict=feeds_train)
            test_acc = sess.run(accr, feed_dict=feeds_test )
            print('epoch %03d/%03d loss:%.9f train_acc:%.3f, test_acc:%.3f'
                  %(epoch, 100,avg_cost,train_acc,test_acc))