剛開始跑的程式碼還有錯誤，但是不知道為什麼再次執行就正確了，可能是新添加了input_data.py檔案，暫存記錄一下，等待深入研究

# -- coding: utf-8 --

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

# 層節點
INPUT_NODE = 784
LAYER1_NODE = 500
OUTPUT_NODE = 10

# 資料batch大小
BATCH_SIZE = 100

# 訓練引數
LEARNING_RATE_BASE = 0.8
LEARNING_RATE_DECAY = 0.99
REGULARIZATION_RATE = 0.0001
TRAINING_STEPS = 30000
MOVING_AVERAGE_DECAY = 0.99


# 前向傳播函式
def inference(input_tensor, avg_class, weights1, biases1, weights2, biases2):
    if avg_class == 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)


def train(mnist):
    # 輸入層和資料label
    x = tf.placeholder(tf.float32, [None, INPUT_NODE], name='x-input')
    y_ = tf.placeholder(tf.float32, [None, OUTPUT_NODE], name='y-input')

    # 隱藏層引數初始化
    weights1 = tf.Variable(tf.truncated_normal([INPUT_NODE, LAYER1_NODE], stddev=0.1))
    biases1 = tf.Variable(tf.constant(0.1, shape=[LAYER1_NODE]))

    # 輸出層引數初始化
    weights2 = tf.Variable(tf.truncated_normal([LAYER1_NODE, OUTPUT_NODE], stddev=0.1))
    biases2 = tf.Variable(tf.constant(0.1, shape=[OUTPUT_NODE]))

    # 前向傳播結果y
    y = inference(x, None, weights1, biases1, weights2, biases2)

    # use for count the train step , trainable=False
    global_step = tf.Variable(0, trainable=False)

    # 滑動平均模型，及加入滑動平均的前向傳播結果average_y
    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)

    # 計算交叉熵，並加入正則-->損失函式loss
    cross_entropy = tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y)
    cross_entropy_mean = tf.reduce_mean(cross_entropy)
    regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
    regularization = regularizer(weights1) + regularizer(weights2)
    loss = cross_entropy_mean + regularization
    # 學習率
    learning_rate = tf.train.exponential_decay(LEARNING_RATE_BASE, global_step, mnist.train.num_examples / BATCH_SIZE,
                                               LEARNING_RATE_DECAY)
    # train_step 梯度下降(學習率，損失函式，全域性步數)
    train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss, global_step=global_step)
    # 運算圖控制，用train_op作集合
    with tf.control_dependencies([train_step, variables_averages_op]):
        train_op = tf.no_op(name='train')
    # 判斷準確率
    correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

    # 持久化
    saver = tf.train.Saver()

    with tf.Session() as sess:
        tf.initialize_all_variables().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):
            # 每1000輪測試一次
            if i % 1000 == 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})
        saver.save(sess, "./model/model.ckpt")
        test_acc = sess.run(accuracy, feed_dict=test_feed)
        print("After %d training step(s), test accuracy using average model is %g" % (TRAINING_STEPS, test_acc))


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


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

# __author__ = 'youngkl'
# -*- coding: utf-8 -*-

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import gzip
import os
import tempfile

import numpy
from six.moves import urllib
from six.moves import xrange # pylint: disable=redefined-builtin
import tensorflow as tf
from tensorflow.contrib.learn.python.learn.datasets.mnist import read_data_sets
 

After 0 training step(s), validation accuracy using average model is 0.0798 

After 1000 training step(s), validation accuracy using average model is 0.9726 

After 2000 training step(s), validation accuracy using average model is 0.9746 

After 3000 training step(s), validation accuracy using average model is 0.982 

After 4000 training step(s), validation accuracy using average model is 0.9796 

After 5000 training step(s), validation accuracy using average model is 0.9822 

After 6000 training step(s), validation accuracy using average model is 0.9832 

After 7000 training step(s), validation accuracy using average model is 0.9854 

After 8000 training step(s), validation accuracy using average model is 0.9824 

After 9000 training step(s), validation accuracy using average model is 0.9846 

After 10000 training step(s), validation accuracy using average model is 0.984 

After 11000 training step(s), validation accuracy using average model is 0.984 

After 12000 training step(s), validation accuracy using average model is 0.9842 

After 13000 training step(s), validation accuracy using average model is 0.9858 

After 14000 training step(s), validation accuracy using average model is 0.984 

After 15000 training step(s), validation accuracy using average model is 0.983 

After 16000 training step(s), validation accuracy using average model is 0.9858 

After 17000 training step(s), validation accuracy using average model is 0.9848 

After 18000 training step(s), validation accuracy using average model is 0.9842 

After 19000 training step(s), validation accuracy using average model is 0.9844 

After 20000 training step(s), validation accuracy using average model is 0.9844 

After 21000 training step(s), validation accuracy using average model is 0.984 

After 22000 training step(s), validation accuracy using average model is 0.9844 

After 23000 training step(s), validation accuracy using average model is 0.9844 

After 24000 training step(s), validation accuracy using average model is 0.984 

After 25000 training step(s), validation accuracy using average model is 0.985 

After 26000 training step(s), validation accuracy using average model is 0.9844 

After 27000 training step(s), validation accuracy using average model is 0.9844 

After 28000 training step(s), validation accuracy using average model is 0.985 

After 29000 training step(s), validation accuracy using average model is 0.9846 

After 30000 training step(s), test accuracy using average model is 0.9843

Process finished with exit code 0