《实战谷歌TensorFlow深度学习》第五章MNIST数字识别

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

INPUT_NODE=784
OUTPUT_NODE=10

LAYER1_NODE=500  #隐藏层的节点数（本结构只有一个隐藏层）

BATCH_SIZE=100

LRARNING_RATE_BASE=0.8
LRARNING_RATE_DECAY=0.99        #学习率的衰减率

REGULARIZATION_RATE=0.0001          #正则化项在损失函数中的系数

TRAINNING_STEPS=30000           #训练轮数
MOVING_AVERAGE_DECAY=0.99       #滑动平均衰减率

def inference(input_tensor,avg_class,weights1,biases1,weights2,biases2):
    if avg_class==None:
        layer1 = tf.nn.relu(tf.matmul(input_tensor,weights1)+biases1)  # 计算隐层的前向传播结果
        return tf.matmul(layer1,weights2)+biases2
    else:
        layer1=tf.nn.relu(tf.matmul(input_tensor,weights1)+avg_class.average(biases1))
        return tf.matmul(layer1,avg_class.average(weights2))+avg_class.average(biases2)



#训练模型过程

def train(mnist):
    x=tf.placeholder(tf.float32,[None,INPUT_NODE],name='x-input')
    y_=tf.placeholder(tf.float32,[None,OUTPUT_NODE],name='y-input')

    #生成隐藏层的参数
    weights1=tf.Variable(tf.truncated_normal([INPUT_NODE,LAYER1_NODE],stddev=0.1))
    biases1=tf.Variable(tf.constant(0.1,shape=[LAYER1_NODE]))

    #生成输出层的数据
    weights2=tf.Variable(tf.truncated_normal([LAYER1_NODE,OUTPUT_NODE],stddev=0.1))
    biases2=tf.Variable(tf.constant(0.1,shape=[OUTPUT_NODE]))

    #计算当前参数前向传播的结果。这里给出的用于计算平均滑动的类为 NONE,所以函数不会使用参数的滑动平均值
    y=inference(x,None,weights1,biases1,weights2,biases2)
    global_step=tf.Variable(0,trainable=False)
    variable_averages=tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY,global_step)
    variables_averages_op=variable_averages.apply(tf.trainable_variables())

    average_y=inference(x,variable_averages,weights1,biases1,weights2,biases2)

    #计算交叉熵                (需要将原文的语句做以下修改)
    cross_entropy=tf.nn.sparse_softmax_cross_entropy_with_logits(labels=tf.arg_max(y_,1) ,logits=y)
    #计算当前batch中所有样例的交叉熵的平均值
    cross_entropy_mean=tf.reduce_mean(cross_entropy)

    #计算L2正则化损失函数
    regularizer=tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
    #计算模型的正则化损失
    regularization=regularizer(weights1)+regularizer(weights2)

    #总损失等于交叉熵损失和正则化损失的和
    loss=cross_entropy_mean + regularization
    #设置指数衰减的学习率
    learning_rate=tf.train.exponential_decay(LRARNING_RATE_BASE,global_step,mnist.train.num_examples/BATCH_SIZE, LRARNING_RATE_DECAY)   #学习率衰减速度
    #优化损失函数，这里的损失函数包括交叉熵损失函数和L2损失函数

    train_step=tf.train.GradientDescentOptimizer(learning_rate).minimize(loss,global_step=global_step)

    #一次进行反向传播参数更新网络中的 参数和参数的滑动平均值
    #train_op=tf.group(train_step,variable_averages_op)     #此句话下面的两句话是等价操作
    with tf.control_dependencies([train_step,variables_averages_op]):
        train_op=tf.no_op(name='train')
    correct_prediction=tf.equal(tf.arg_max(average_y,1),tf.arg_max(y_,1))
    accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))

    #初始化会话并开始训练过程
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())             #初始化所有的变量
        validate_feed={x : mnist.validation.images,
                       y_ :mnist.validation.labels}
        test_feed={x : mnist.test.images, y_ : mnist.test.labels}

        #迭代的训练神经网络
        for i in range(TRAINNING_STEPS):
            if i % 1000 == 0:
                validate_acc=sess.run(accuracy,feed_dict=validate_feed)
                print("After %d trainingstep(s), validation accuracy using average model is %g"% (i,validate_acc))

            xs,ys=mnist.train.next_batch(BATCH_SIZE)
            sess.run(train_op,feed_dict={x:xs,y_:ys})
        test_acc=sess.run(accuracy,feed_dict=test_feed)
        print("After %d training step(s),test accuracy using average model is %g"%(TRAINNING_STEPS,test_acc))


#注程序入口

def main(argv=None):
    mnist=input_data.read_data_sets("/tmp/data/",one_hot=True)
    train(mnist)


if __name__=="__main__":
    tf.app.run()

在TensorFlow 1.5版本中执行如下所示：

I:\SoftwareLe\Anaconda\envs\tf15py35\python.exe D:/pydemo/pythonlLearning/deep_learning/mnist_demo.py
Extracting /tmp/data/train-images-idx3-ubyte.gz
Extracting /tmp/data/train-labels-idx1-ubyte.gz
Extracting /tmp/data/t10k-images-idx3-ubyte.gz
Extracting /tmp/data/t10k-labels-idx1-ubyte.gz

After 0 trainingstep(s), validation accuracy using average model is 0.1064
After 1000 trainingstep(s), validation accuracy using average model is 0.9726
After 2000 trainingstep(s), validation accuracy using average model is 0.9786
After 3000 trainingstep(s), validation accuracy using average model is 0.9796
After 4000 trainingstep(s), validation accuracy using average model is 0.9816
After 5000 trainingstep(s), validation accuracy using average model is 0.9824
After 6000 trainingstep(s), validation accuracy using average model is 0.9822
After 7000 trainingstep(s), validation accuracy using average model is 0.9842
After 8000 trainingstep(s), validation accuracy using average model is 0.9836
After 9000 trainingstep(s), validation accuracy using average model is 0.9834
After 10000 trainingstep(s), validation accuracy using average model is 0.9832
After 11000 trainingstep(s), validation accuracy using average model is 0.9838
After 12000 trainingstep(s), validation accuracy using average model is 0.985
After 13000 trainingstep(s), validation accuracy using average model is 0.9828
After 14000 trainingstep(s), validation accuracy using average model is 0.985
After 15000 trainingstep(s), validation accuracy using average model is 0.9836
After 16000 trainingstep(s), validation accuracy using average model is 0.9846
After 17000 trainingstep(s), validation accuracy using average model is 0.9842
After 18000 trainingstep(s), validation accuracy using average model is 0.9838
After 19000 trainingstep(s), validation accuracy using average model is 0.9844
After 20000 trainingstep(s), validation accuracy using average model is 0.9854
After 21000 trainingstep(s), validation accuracy using average model is 0.985
After 22000 trainingstep(s), validation accuracy using average model is 0.9838
After 23000 trainingstep(s), validation accuracy using average model is 0.9844
After 24000 trainingstep(s), validation accuracy using average model is 0.9842
After 25000 trainingstep(s), validation accuracy using average model is 0.9848
After 26000 trainingstep(s), validation accuracy using average model is 0.9844
After 27000 trainingstep(s), validation accuracy using average model is 0.9852
After 28000 trainingstep(s), validation accuracy using average model is 0.9852
After 29000 trainingstep(s), validation accuracy using average model is 0.9858
After 30000 training step(s),test accuracy using average model is 0.9829

Process finished with exit code 0