训练卷积神经网络识别mnist数字

总的来说，思路较为清晰，关键搞清卷积过程以及过程中张量维度的变化，还需注意求正确率的方法——利用平均值求解

from  tensorflow.examples.tutorials.mnist  import  input_data

import tensorflow as tf

#随机化权值变量tensor，高斯分布

def  weight_variable(shape):

             initial=tf.truncated_normal(shape,stddev=0.1)

             return  tf.Variable(initial)

#随机化偏置，高斯分布

def  bias_variable(shape):

           initial=tf.constant(0.1,shape=shape)

           return  tf.Variable(initial)

#定义二维图像卷积

def  conv2d(x,W):

        return    tf.nn.conv2d(x,W,strides=[1,1,1,1],padding='SAME')

def   max_pool_2x2(x):

         return  tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

###start here!###

sess=tf.InteractiveSession()

mnist=input_data.read_data_sets('MNIST_data',one_hot=True)

#接收mnist中真实数据

x=tf.placeholder("float",shape=[None,784])

y_=tf.placeholder("float",shape=[None,10])

#layer 1: convolution + relu + max pooling

W_conv1=weight_variable([5,5,1,32])

b_conv1=bias_variable([32])

x_image=tf.reshape(x,[-1,28,28,1])#[batch, height, width, channels]

h_conv1=tf.nn.relu(conv2d(x_image,W_conv1)+b_conv1)

h_pool1=max_pool_2x2(h_conv1)

#layer 2: convolution + relu + max pooling

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)

W_fc1=weight_variable([7*7*64,1024])

b_fc1=bias_variable([1024])

#第三层 全连接层

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)

#dropout层

keep_prob=tf.placeholder("float")

h_fc1_drop=tf.nn.dropout(h_fc1,keep_prob)

#全连接层

W_fc2=weight_variable([1024,10])

b_fc2=bias_variable([10])

#softmax 判定层

y_conv=tf.nn.softmax(tf.matmul(h_fc1_drop,W_fc2)+b_fc2)

cross_entropy= -tf.reduce_sum(y_*tf.log(y_conv))#交叉熵cost计算方法

train_step=tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)#ada优化

correct_prediction=tf.equal(tf.arg_max(y_conv,1),tf.arg_max(y_,1))

accuracy=tf.reduce_mean(tf.cast(correct_prediction,"float"))

sess.run(tf.global_variables_initializer())

for  i  inrange(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})

print("test accuracy %g"%accuracy.eval(feed_dict={x:mnist.test.images,y_:mnist.test.labels,keep_prob:1.0}))