import tensorflow as tf# 输入数据from tensorflow.examples.tutorials.mnist import input_datamnist = input_data.read_data_sets("E:\\MNIST_data", one_hot=True)# 定义网络的超参数learning_rate = 0.001training_iters = 200000batch_size = 128display_step = 5# 定义网络的参数# 输入的维度 (img shape: 28*28)n_input = 784 # 标记的维度 (0-9 digits)n_classes = 10 # Dropout的概率，输出的可能性dropout = 0.75 # 输入占位符x = tf.placeholder(tf.float32, [None, n_input])y = tf.placeholder(tf.float32, [None, n_classes])#dropout (keep probability)keep_prob = tf.placeholder(tf.float32) # 定义卷积操作def conv2d(name,x, W, b, strides=1):# Conv2D wrapper, with bias and relu activationx = tf.nn.conv2d(x, W, strides=[1, strides, strides, 1], padding='SAME')x = tf.nn.bias_add(x, b)# 使用relu激活函数return tf.nn.relu(x,name=name) # 定义池化层操作def maxpool2d(name,x, k=2):# MaxPool2D wrapperreturn tf.nn.max_pool(x, ksize=[1, k, k, 1], strides=[1, k, k, 1],padding='SAME',name=name)# 规范化操作def norm(name, l_input, lsize=4):return tf.nn.lrn(l_input, lsize, bias=1.0, alpha=0.001 / 9.0,beta=0.75, name=name)# 定义所有的网络参数weights = {'wc1': tf.Variable(tf.random_normal([11, 11, 1, 96])),'wc2': tf.Variable(tf.random_normal([5, 5, 96, 256])),'wc3': tf.Variable(tf.random_normal([3, 3, 256, 384])),'wc4': tf.Variable(tf.random_normal([3, 3, 384, 384])),'wc5': tf.Variable(tf.random_normal([3, 3, 384, 256])),'wd1': tf.Variable(tf.random_normal([4*4*256, 4096])),'wd2': tf.Variable(tf.random_normal([4096, 1024])),'out': tf.Variable(tf.random_normal([1024, n_classes]))}biases = {'bc1': tf.Variable(tf.random_normal([96])),'bc2': tf.Variable(tf.random_normal([256])),'bc3': tf.Variable(tf.random_normal([384])),'bc4': tf.Variable(tf.random_normal([384])),'bc5': tf.Variable(tf.random_normal([256])),'bd1': tf.Variable(tf.random_normal([4096])),'bd2': tf.Variable(tf.random_normal([1024])),'out': tf.Variable(tf.random_normal([n_classes]))}# 定义整个网络def alex_net(x, weights, biases, dropout):# 向量转为矩阵 Reshape input picturex = tf.reshape(x, shape=[-1, 28, 28, 1])# 第一层卷积# 卷积conv1 = conv2d('conv1', x, weights['wc1'], biases['bc1'])# 下采样pool1 = maxpool2d('pool1', conv1, k=2)# 规范化norm1 = norm('norm1', pool1, lsize=4)# 第二层卷积# 卷积conv2 = conv2d('conv2', norm1, weights['wc2'], biases['bc2'])# 最大池化（向下采样）pool2 = maxpool2d('pool2', conv2, k=2)# 规范化norm2 = norm('norm2', pool2, lsize=4)# 第三层卷积# 卷积conv3 = conv2d('conv3', norm2, weights['wc3'], biases['bc3'])# 规范化norm3 = norm('norm3', conv3, lsize=4)# 第四层卷积conv4 = conv2d('conv4', norm3, weights['wc4'], biases['bc4'])# 第五层卷积conv5 = conv2d('conv5', conv4, weights['wc5'], biases['bc5'])# 最大池化（向下采样）pool5 = maxpool2d('pool5', conv5, k=2)# 规范化norm5 = norm('norm5', pool5, lsize=4)# 全连接层1fc1 = tf.reshape(norm5, [-1, weights['wd1'].get_shape().as_list()[0]])fc1 =tf.add(tf.matmul(fc1, weights['wd1']),biases['bd1'])fc1 = tf.nn.relu(fc1)# dropoutfc1=tf.nn.dropout(fc1,dropout)# 全连接层2fc2 = tf.reshape(fc1, [-1, weights['wd2'].get_shape().as_list()[0]])fc2 =tf.add(tf.matmul(fc2, weights['wd2']),biases['bd2'])fc2 = tf.nn.relu(fc2)# dropoutfc2=tf.nn.dropout(fc2,dropout)# 输出层out = tf.add(tf.matmul(fc2, weights['out']) ,biases['out'])return out# 构建模型pred = alex_net(x, weights, biases, keep_prob)# 定义损失函数和优化器cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred, labels=y))optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)# 评估函数correct_pred = tf.equal(tf.argmax(pred,1), tf.argmax(y,1))accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))# 初始化变量init = tf.global_variables_initializer()# 开启一个训练with tf.Session() as sess:sess.run(init)step = 1# 开始训练，直到达到training_iters，即200000while step * batch_size < training_iters:#获取批量数据batch_x, batch_y = mnist.train.next_batch(batch_size)sess.run(optimizer, feed_dict={x: batch_x, y: batch_y, keep_prob: dropout})if step % display_step == 0:# 计算损失值和准确度，输出loss,acc = sess.run([cost,accuracy], feed_dict={x: batch_x, y: batch_y, keep_prob: 1.})print ("Iter " + str(step*batch_size) + ", Minibatch Loss= " + "{:.6f}".format(loss) + ", Training Accuracy= " + "{:.5f}".format(acc))step += 1print ("Optimization Finished!")# 计算测试集的精确度print ("Testing Accuracy:",sess.run(accuracy, feed_dict={x: mnist.test.images[:256],y: mnist.test.labels[:256],keep_prob: 1.}))
