为什么计算loss的时候用的是未经过滑动平均的y， 而计算精度的时候用的是经过滑动平均的average_y ？

Question

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

# MNIST数据集相关的常数
INPUT_NODE = 784  # 输入层的节点数
OUTPUT_NODE = 10  # 输出层的节点数

# 配置神经网络的参数
LAYER1_NODE = 500  # 隐藏层的节点数
BATCH_SIZE = 100  # 批量大小

LEARNING_RATE_BASE = 0.8  # 基础的学习率
LEARNING_RATE_DECAY = 0.99  # 学习率的衰减率
REGULARIZATION_RATE = 0.0001  # 描述模型复杂度的正则化项在损失函数中的系数
TRAINING_STEPS = 30000
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 is None:
        layer1 = tf.nn.relu(tf.matmul(input_tensor, weights1) + biases1)

        return tf.matmul(layer1, weights2) + biases2
    else:
        layer1 = tf.nn.relu(tf.matmul(input_tensor, avg_class.average(weights1)) + avg_class.average(biases1))

        return tf.matmul(layer1, avg_class.average(weights2)) + avg_class.average(biases2)


# 训练模型的过程
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]))

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

    # 计算在当前参数下神经网络前向传播的结果
    y = inference(x, None, weights1, biases1, weights2, biases2)

    # 定义存储训练轮数的变量。一般指定为不可训练的参数
    global_step = tf.Variable(0, trainable=False)

    # 给定滑动平均衰减率和训练轮数的变量，初始化滑动平均类
    variable_averages = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)

    # 在所有代表神经网络参数的变量上使用滑动平均。其他辅助变量不需要。tf.trainable_variables()返回的是图上集合
    variables_averages_op = variable_averages.apply(tf.trainable_variables())

    # 计算使用了滑动平均后的前向传播结果。
    average_y = inference(x, variable_averages, weights1, biases1, weights2, biases2)

    # 计算交叉熵作为刻画预测值和真实值之间差距的损失函数。 这里使用了sparse_softmax_cross_entropy_with_logits函数来计算交叉熵。
    # tf.argmax()返回的是最大值的索引号
    cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y, labels=tf.argmax(y_, 1))
    # 计算在当前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(
        LEARNING_RATE_BASE,  # 基础的学习率，随着迭代的进行，更新变量时使用
        global_step,  # 当前迭代的轮数
        mnist.train.num_examples / BATCH_SIZE,  # 过完所有的训练数据需要的迭代次数
        LEARNING_RATE_DECAY,  # 学习率衰减速度
    )

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

    # 反向传播更新神经网络的参数， 以及更新每一个参数的滑动平均值
    train_op = tf.group(train_step, variables_averages_op)

    # 书上写的是correct_prediction = tf.equal(tf.argmax(average_y, 1), tf.argmax(y_, 1))
    correct_prediction = tf.equal(tf.argmax(average_y, 1), tf.argmax(y_, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

    # 初始化会话并开始训练的过程
    with tf.Session() as sess:
        tf.global_variables_initializer().run()
        validate_feed = {x: mnist.validation.images,
                         y_: mnist.validation.labels}

        test_feed = {x: mnist.test.images,
                     y_: mnist.test.labels}

        for i in range(TRAINING_STEPS):
            if i % 5000 == 0:
                validate_acc = sess.run(accuracy, feed_dict=validate_feed)
                print("After %d training step(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 models is %g " % (TRAINING_STEPS, test_acc))


def main(argv=None):
    mnist = input_data.read_data_sets("./path/to/MNIST_data", one_hot=True)
    train(mnist)


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