TensorFlow学习笔记-实现经典LeNet5模型

　　LeNet5模型是Yann LeCun教授于1998年提出来的，它是第一个成功应用于数字识别问题的卷积神经网络。在MNIST数据中，它的准确率达到大约99.2%.
　　通过TensorFlow实现的LeNet5模型，主要用到在说使用变量管理，可以增加代码可读性、降低代码冗余量，提高编程效率，更方便管理变量。我们将LeNet5模型分为三部分：
　　1、网络定义部分：这部分是训练和验证都需要的网络结构。
　　2、训练部分：用于神经网络训练MNIST训练集。
　　3、验证部分：验证训练模型的准确率，在Tensorflow训练过程中，可以实时验证模型的正确率。
　　将训练部分与验证部分分开的好处在于，训练部分可以持续输出训练好的模型，验证部分可以每隔一段时间验证模型的准确率；如果模型不好，则需要及时调整网络结构的参数。

一、 网络定义部分

import tensorflow as tfINPUT_NODE = 784OUTPUT_NODE = 10IMAGE_SIZE = 28NUM_CHANNEL = 1NUM_LABEL = 10# LAYER1CONV1_DEEP = 32CONV1_SIZE = 5# LAYER2CONV2_DEEP = 64CONV2_SIZE = 5# 全连接层FC_SIZE = 512# LAYER1_NODE = 500def interence(input_tensor,train,regularizer):    with tf.variable_scope('layer1-conv'):        w = tf.get_variable('w', [CONV1_SIZE,CONV1_SIZE,NUM_CHANNEL,CONV1_DEEP],                            initializer=tf.truncated_normal_initializer(stddev=0.1))        b = tf.get_variable('b',shape=[CONV1_DEEP],initializer=tf.constant_initializer(0.0))        # filter shape is :[filter_height, filter_width, in_channels, out_channels]        # input tensor shape is:[batch, in_height, in_width, in_channels]        # `strides = [1, stride, stride, 1]`.        # return [batch, height, width, channels].        conv1 = tf.nn.conv2d(input_tensor,w,strides=[1,1,1,1],padding='SAME')        relu1 = tf.nn.relu(tf.nn.bias_add(conv1,b))    with tf.variable_scope('layer2-pool'):        pool1 = tf.nn.max_pool(relu1,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')    with tf.variable_scope('layer3-conv'):        w = tf.get_variable('w', [CONV2_SIZE, CONV2_SIZE, CONV1_DEEP, CONV2_DEEP],                            initializer=tf.truncated_normal_initializer(stddev=0.1))        b = tf.get_variable('b',shape=[CONV2_DEEP],initializer=tf.constant_initializer(0.0))        conv2 = tf.nn.conv2d(pool1, w, strides=[1, 1, 1, 1], padding='SAME')        relu2 = tf.nn.relu(tf.nn.bias_add(conv2, b))    with tf.variable_scope('layer4-pool'):        # pool2 size is [batch_size,7,7,64]        pool2 = tf.nn.max_pool(relu2,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')    # 接下来是全连接层，需要将pool2转换为一维向量，作为后面的输入    pool_shape = pool2.get_shape().as_list()    nodes = pool_shape[1] * pool_shape[2] * pool_shape[3]    reshaped = tf.reshape(pool2,[-1,nodes])    # reshaped = tf.reshape(pool2,[BATCH_SIZE,-1])    # print(reshaped.get_shape())    with tf.variable_scope('layer5-fc1'):        fc1_w = tf.get_variable('w',shape=[nodes,FC_SIZE],initializer=tf.truncated_normal_initializer(stddev=0.1))        try:            # 只有全连接层的权重需要加入正则化            if regularizer != None:                tf.add_to_collection('loss',regularizer(fc1_w))        except:            pass        fc1_b = tf.get_variable('b',shape=[FC_SIZE],initializer=tf.constant_initializer(0.1))        fc1 = tf.nn.relu(tf.matmul(reshaped,fc1_w) + fc1_b)        # 使用Dropout随机将部分节点的输出改为0，为了防止过拟合的现象，从而使模型在测试数据中表现更好。        # dropout一般只会在全连接层使用。        if train:            fc1 = tf.nn.dropout(fc1,0.5)    with tf.variable_scope('layer6-fc2'):        fc2_w = tf.get_variable('w', shape=[FC_SIZE, NUM_LABEL], initializer=tf.truncated_normal_initializer(stddev=0.1))        try:            if regularizer != None:                tf.add_to_collection('loss', regularizer(fc2_w))        except:            pass        fc2_b = tf.get_variable('b', shape=[NUM_LABEL], initializer=tf.constant_initializer(0.1))        # 最后一层的输出，不需要加入激活函数        logit = tf.matmul(fc1, fc2_w) + fc2_b    return logit

二、训练部分

import osimport tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_datafrom mnist_cnn import mnist_interenceimport numpy as npBATCH_SIZE = 100LEARNING_RATE_BASE = 0.8LEARNING_RATE_DECAY = 0.99REGULARIZATION_TATE = 0.0001MOVING_AVERAGE_DECAY = 0.99TRAIN_STEP = 300000MODEL_PATH = 'model'MODEL_NAME = 'model'def train(mnist):    x = tf.placeholder(tf.float32, shape=[None,                                          mnist_interence.IMAGE_SIZE,                                          mnist_interence.IMAGE_SIZE,                                          mnist_interence.NUM_CHANNEL ], name='x-input')    y_ = tf.placeholder(tf.float32, shape=[None, mnist_interence.OUTPUT_NODE], name='y-input')    regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_TATE)    y = mnist_interence.interence(x,True,regularizer)    global_step = tf.Variable(0, trainable=False)    variable_average = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)    variable_average_ops = variable_average.apply(tf.trainable_variables())    cross_entroy = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y, labels=tf.argmax(y_, 1))    cross_entroy_mean = tf.reduce_mean(cross_entroy)    loss = cross_entroy_mean + tf.add_n(tf.get_collection('loss'))    learning_rate = tf.train.exponential_decay(LEARNING_RATE_BASE, global_step,                                               mnist.train.num_examples / BATCH_SIZE, LEARNING_RATE_DECAY)    train_step = tf.train.GradientDescentOptimizer(0.01).minimize(loss, global_step=global_step)    train_op = tf.group(train_step, variable_average_ops)    saver = tf.train.Saver()    with tf.Session() as sess:        tf.global_variables_initializer().run()        for i in range(TRAIN_STEP):            # 由于神经网络的输入大小为[BATCH_SIZE,IMAGE_SIZE,IMAGE_SIZE,CHANNEL]，因此需要reshape输入。            xs,ys = mnist.train.next_batch(BATCH_SIZE)            reshape_xs = np.reshape(xs,(BATCH_SIZE, mnist_interence.IMAGE_SIZE,                                        mnist_interence.IMAGE_SIZE,                                        mnist_interence.NUM_CHANNEL))            # print(type(xs))            _,loss_value,step,learn_rate = sess.run([train_op,loss,global_step,learning_rate],feed_dict={x:reshape_xs,y_:ys})            if i % 1000 == 0:                print('After %d step, loss on train is %g,and learn rate is %g'%(step,loss_value,learn_rate))                saver.save(sess,os.path.join(MODEL_PATH,MODEL_NAME),global_step=global_step)def main():    mnist = input_data.read_data_sets('../mni_data', one_hot=True)    # ys = mnist.validation.labels    # print(ys)    train(mnist)if __name__ == '__main__':    main()

验证部分

import tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_datafrom mnist_cnn import mnist_interencefrom mnist_cnn import mnist_trainEVAL_INTERVAL_SECS = 10BATCH_SIZE = 100import timeimport numpy as npdef evaluate(mnist):    with tf.Graph().as_default():        x = tf.placeholder(tf.float32, shape=[None,                                              mnist_interence.IMAGE_SIZE,                                              mnist_interence.IMAGE_SIZE,                                              mnist_interence.NUM_CHANNEL], name='x-input')        y_ = tf.placeholder(tf.float32, shape=[None,mnist_interence.OUTPUT_NODE], name='y-input')        xs, ys = mnist.validation.images, mnist.validation.labels        reshape_xs = np.reshape(xs, (-1, mnist_interence.IMAGE_SIZE,                                     mnist_interence.IMAGE_SIZE,                                     mnist_interence.NUM_CHANNEL))        print(mnist.validation.labels[0])        val_feed = {x: reshape_xs, y_: mnist.validation.labels}        y = mnist_interence.interence(x,False,None)        correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))        accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))        variable_average = tf.train.ExponentialMovingAverage(mnist_train.MOVING_AVERAGE_DECAY)        val_to_restore = variable_average.variables_to_restore()        saver = tf.train.Saver(val_to_restore)        while True:            with tf.Session() as sess:                ckpt = tf.train.get_checkpoint_state(mnist_train.MODEL_PATH)                if ckpt and ckpt.model_checkpoint_path:                    saver.restore(sess,ckpt.model_checkpoint_path)                    global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]                    accuracy_score = sess.run(accuracy,feed_dict=val_feed)                    print('After %s train ,the accuracy is %g'%(global_step,accuracy_score))                else:                    print('No Checkpoint file find')                    # continue            time.sleep(EVAL_INTERVAL_SECS)def main():    mnist = input_data.read_data_sets('../mni_data',one_hot=True)    evaluate(mnist)if __name__ == '__main__':    main()

　　最后，在MNIST数据集中的准确率大约在99.4%左右