Tensorflow实现简单的卷积神经网络

网络基本参数及架构

X ：[None,784]
X_image :[None,28,28,1]
y_:[None,10]

Conv1 : [None,28,28,32]
pool1 : [None,14,14,32]

Conv2 : [None,14,14,64]
pool2 : [None,7,7,64]

h_pool2_flat : [None,7764]
h_fc1 : [None,1024]
y_conv : [None,10]

第一个卷积：卷积核55 通道数=1 卷积核个数 = 32个
第一个池化：22–>1最大池化
第二个卷积：卷积核55 通道数=32 卷积核个数 = 64个
第二个池化：22–>1最大池化

程序如下

 - `# -*- coding: utf-8 -*-
"""
Created on Mon Jul 22 15:28:17 2019
@author: ADMIN
"""

 
from tensorflow.examples.tutorials.mnist import input_data 
import tensorflow as tf 
mnist = input_data.read_data_sets("C:\\Users\\ADMIN\\Desktop\\shuju", one_hot=True) 
sess = tf.InteractiveSession() 
 
 
#截断正态分布  标准差=0.1
def weight_variable(shape): 
    initial = tf.truncated_normal(shape, stddev=0.1)  
    return tf.Variable(initial) 
 
#bias初始化值0.1. 
def bias_variable(shape): 
    initial = tf.constant(0.1, shape=shape) #bias=0.1
    return tf.Variable(initial) 
 
 
def conv2d(x, W): 
  return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME') 
 

#strides:[1,2,2,1]表示横竖方向步长为2 
def max_pool_2x2(x): 
  return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides = [1, 2, 2, 1], padding='SAME') 
                            #表示1个2*2的卷积核 通道数为1
 
x = tf.placeholder(tf.float32, [None, 784]) 
y_ = tf.placeholder(tf.float32, [None, 10]) 
x_image = tf.reshape(x, [-1, 28, 28, 1]) 
#-1：样本数量不固定 28,28：新形状的shape 1:颜色通道数 
#即把原来的1*784变为28*28，通道数为1

#[5,5,1,32]：卷积核的尺寸为 5×5, 颜色通道为 1, 卷积核个数32个
W_conv1 = weight_variable([5, 5, 1, 32]) 
b_conv1 = bias_variable([32]) 
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1) 
h_pool1 = max_pool_2x2(h_conv1) 
 
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) 
 
#因为padding='SAME'  两次卷积不改变大小   
#2次2×2的池化后，图像的尺寸变为7×7 通道数为64  卷积最后输出为7×7×64. 

#tensor进入全连接层之前，先将64张二维图像变形为1维图像，便于计算。 
W_fc1 = weight_variable([7*7*64, 1024]) 
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) 
 
 
#对全连接层做dropot   防止overfiting
keep_prob = tf.placeholder(tf.float32) 
h_fc1_dropout = tf.nn.dropout(h_fc1, keep_prob) 
 
 
#又一个全连接后softmax分类 
W_fc2 = weight_variable([1024, 10]) 
b_fc2 = bias_variable([10]) 
y_conv = tf.nn.softmax(tf.matmul(h_fc1_dropout, W_fc2) + b_fc2) 
 
 
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_*tf.log(y_conv), reduction_indices=[1])) 
#AdamOptimizer：Adam优化函数 
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy) 
 
 
 
correct_prediction = tf.equal(tf.argmax(y_, 1), tf.argmax(y_conv, 1)) 
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32)) 
 
 
#训练，并且每100个batch计算一次精度 
tf.global_variables_initializer().run() 
for i in range(20000): 
  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}) 
 
 
#在测试集上测试 print("test accuracy %g"%accuracy.eval(feed_dict={x:mnist.test.images, y_:mnist.test.labels, `