先弄懂卷积神经网络的原理,推荐这两篇博客:http://blog.csdn.net/yunpiao123456/article/details/52437794 http://blog.csdn.net/qq_25762497/article/details/51052861#%E6%A6%82%E6%8F%BD
简单的测试程序如下(具体各参数代表什么可以百度):
1 from tensorflow.examples.tutorials.mnist import input_data 2 import tensorflow as tf 3 4 mnist=input_data.read_data_sets("MNIST_data/",one_hot=True) 5 sess=tf.InteractiveSession() 6 7 def weight_variable(shape): 8 initial=tf.truncated_normal(shape,stddev=0.1) 9 return tf.Variable(initial) 10 11 def bias_variable(shape): 12 initial=tf.constant(0.1,shape=shape) 13 return tf.Variable(initial) 14 15 def conv2d(x,w): 16 return tf.nn.conv2d(x,w,strides=[1,1,1,1],padding=‘SAME‘) 17 18 def max_pool_2x2(x): 19 return tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding=‘SAME‘) 20 21 x=tf.placeholder(tf.float32,[None,784]) 22 y_=tf.placeholder(tf.float32,[None,10]) 23 x_image=tf.reshape(x,[-1,28,28,1]) 24 25 w_conv1=weight_variable([5,5,1,32]) 26 b_conv1=bias_variable([32]) 27 h_conv1=tf.nn.relu(conv2d(x_image,w_conv1)+b_conv1) 28 h_pool1=max_pool_2x2(h_conv1) 29 30 w_conv2=weight_variable([5,5,32,64]) 31 b_conv2=bias_variable([64]) 32 h_conv2=tf.nn.relu(conv2d(h_pool1,w_conv2)+b_conv2) 33 h_pool2=max_pool_2x2(h_conv2) 34 35 w_fc1=weight_variable([7*7*64,1024]) 36 b_fc1=bias_variable([1024]) 37 h_pool2_flat=tf.reshape(h_pool2,[-1,7*7*64]) 38 h_fc1=tf.nn.relu(tf.matmul(h_pool2_flat,w_fc1)+b_fc1) 39 40 keep_prob=tf.placeholder(tf.float32) 41 h_fc1_drop=tf.nn.dropout(h_fc1,keep_prob) 42 43 w_fc2=weight_variable([1024,10]) 44 b_fc2=bias_variable([10]) 45 y_conv=tf.nn.softmax(tf.matmul(h_fc1_drop,w_fc2)+b_fc2) 46 47 cross_entropy=tf.reduce_mean(-tf.reduce_sum(y_*tf.log(y_conv),reduction_indices=[1])) 48 train_step=tf.train.AdamOptimizer(1e-4).minimize(cross_entropy) 49 50 correct_prediction=tf.equal(tf.argmax(y_conv,1),tf.argmax(y_,1)) 51 accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32)) 52 53 tf.initialize_all_variables().run() 54 for i in range(20000): 55 batch=mnist.train.next_batch(50) 56 if i%100==0: 57 train_accuracy=accuracy.eval(feed_dict={x:batch[0],y_:batch[1],keep_prob:1.0}) 58 print("step %d,training accuracy %g"%(i,train_accuracy)) 59 train_step.run(feed_dict={x:batch[0],y_:batch[1],keep_prob:0.5}) 60 61 print("test accuracy %g"%accuracy.eval(feed_dict={x:mnist.test.images,y_:mnist.test.labels,keep_prob:1.0}))
运行结果: