tensorflow3-手写数字识别

网络结构：

28*28个输入单元，200个中间单元，10个输出单元

代码：

# -*- coding: utf-8 -*-
"""
Created on Fri May 17 19:39:39 2019

@author: 666
"""
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

#输入数据集
mnist = input_data.read_data_sets("F:/AI/AItest/MNIST_data",one_hot=True)

#定义每个批次的大小
batch_size = 100
#计算共有多少个批次
n_batch = mnist.train.num_examples 

#定义两个placeholder
x = tf.placeholder(tf.float32,[None,784])
y = tf.placeholder(tf.float32,[None,10])

#创建简单神经网络
#输入784神经元，中间层为200个单元，输出10个神经元
#定义中间层
W1 = tf.Variable(tf.zeros([784,200]))
b1 = tf.Variable(tf.zeros([200]))
L1 = tf.nn.relu(tf.matmul(x,W1) + b1)

#定义输出层
W =tf.Variable(tf.zeros([200,10]))
b = tf.Variable(tf.zeros([10]))
prediction = tf.nn.softmax(tf.matmul(L1,W)+b)

# 二次代价函数
loss = tf.reduce_mean(tf.square(y-prediction))
#使用梯度下降法
train_step = tf.train.GradientDescentOptimizer(0.2).minimize(loss)

#初始化变量
init = tf.global_variables_initializer()

#判断预测值和正确的值是否一样
#tf.argmax（a，b），a是一个一维张量，1是维度，就是返回a的最大值
correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(prediction,1))

#求准确率 cast是转换数据类型，将bool转成float32
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))


with tf.Session() as sess:
    sess.run(init)
    for epoch in range(21):
        for batch in range(n_batch):
#batch_xs保存数据，batch_ys保存标签
            batch_xs,batch_ys = mnist.train.next_batch(batch_size)
            sess.run(train_step,feed_dict={x:batch_xs,y:batch_ys})
            
        acc = sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels})            
        print("Iter " + str(epoch) +", Testing Accuracy "+ str(acc))

仅仅是为了熟悉如何构建网络的代码，忽略准确率。

改进：更改了变量初始化的方式，使用了dropout

# -*- coding: utf-8 -*-
"""
Created on Tue May 21 19:10:24 2019

@author: miaow
"""

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

#输入数据集
mnist = input_data.read_data_sets("F:/AI/AItest/MNIST_data",one_hot=True)

#定义每个批次的大小
batch_size = 100
#计算共有多少个批次
n_batch = mnist.train.num_examples 

#定义两个placeholder
x = tf.placeholder(tf.float32,[None,784])
y = tf.placeholder(tf.float32,[None,10])
keep_prob = tf.placeholder(tf.float32)

#定义第一层
W1 =tf.Variable(tf.truncated_normal([784,2000],stddev=0.1))
b1 = tf.Variable(tf.zeros([2000]+0.1))
L1 = tf.nn.tanh(tf.matmul(x,W1)+b1)
L1_drop = tf.nn.dropout(L1,keep_prob)


#定义第er层
W2 =tf.Variable(tf.truncated_normal([2000,2000],stddev=0.1))
b2 = tf.Variable(tf.zeros([2000]+0.1))
L2 = tf.nn.tanh(tf.matmul(L1_drop,W2)+b2)
L2_drop = tf.nn.dropout(L2,keep_prob)

#定义输出层
W3 =tf.Variable(tf.zeros([2000,10]))
b3 = tf.Variable(tf.zeros([10]))
prediction = tf.nn.softmax(tf.matmul(L2_drop,W)+b)

# 二次代价函数
loss = tf.reduce_mean(tf.square(y-prediction))
#使用梯度下降法
train_step = tf.train.GradientDescentOptimizer(0.2).minimize(loss)

#初始化变量
init = tf.global_variables_initializer()

#判断预测值和正确的值是否一样
#tf.argmax（a，b），a是一个一维张量，1是维度，就是返回a的最大值
correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(prediction,1))

#求准确率 cast是转换数据类型，将bool转成float32
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))


with tf.Session() as sess:
    sess.run(init)
    for epoch in range(21):
        for batch in range(n_batch):
#batch_xs保存数据，batch_ys保存标签
            batch_xs,batch_ys = mnist.train.next_batch(batch_size)
            sess.run(train_step,feed_dict={x:batch_xs,y:batch_ys,keep_prob:1.0})
            
        test_acc = sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels})
        train_acc = sess.run(accuracy,feed_dict={x:mnist.rain.images,y:mnist.train.labels})    
        print("Iter " + str(epoch) +", Testing Accuracy "+ str(test_acc)+", Training Accuracy "+ str(train_acc))