MNIST数字识别问题（Tensorflow）

一、首先介绍tensorflow持久化的工作原理，持久化代码实现。

1.1使用tf.train.Saver类,以下代码给出了保存tensorflow计算图的方法。

import tensorflow as tf
v1=tf.Variable(tf.constant(1.0,shape=[1]),name="v1")
v2=tf.Variable(tf.constant(2.0,shape=[1]),name="v2")
result=v1+v2
init_op=tf.initialize_all_variables()
saver=tf.train.Saver()
with tf.Session() as sess:
    sess.run(init_op)
    saver.save(sess,"D:/gj20170720/model.ckpt")

1.2加载已保存的模型的方法

import tensorflow as tf
__author__ = 'casgj'
v1=tf.Variable(tf.constant(1.0,shape=[1]),name="v1")
v2=tf.Variable(tf.constant(2.0,shape=[1]),name="v2")
result=v1 + v2
saver=tf.train.Saver()
with tf.Session() as sess:
    saver.restore(sess,"D:/gj20170720/model.ckpt")
    print(sess.run(result))

如果不希望重复定义图上的运算，也可以：

import tensorflow as tf
__author__ = 'casgj'

saver=tf.train.import_meta_graph(("D:/gj20170720/model.ckpt.meta"))

with tf.Session() as sess:
    saver.restore(sess,"D:/gj20170720/model.ckpt")
    print(sess.run

1.3变量重命名的使用

__author__ = 'casgj'
import tensorflow as tf
v1=tf.Variable(tf.constant(1.0,shape=[1]),name="other-v1")
v2=tf.Variable(tf.constant(2.0,shape=[1]),name="other-v2")
saver=tf.train.Saver({
   "v1":v1,"v2":v2})

1.4保存滑动平均模型的运用

_author__ = 'casgj'
import tensorflow as tf
v=tf.Variable(0,dtype=tf.float32,name="v")
for variables in tf.all_variables():
    print(variables.name)

ema=tf.train.ExponentialMovingAverage(0.99)
maintain_averages_op=ema.apply(tf.all_variables())


for variables in tf.all_variables():
    print(variables.name)


saver=tf.train.Saver()
with tf.Session() as sess:
    init_op=tf.initialize_all_variables()
    sess.run(init_op)

    sess.run(tf.assign(v,10))
    sess.run(maintain_averages_op)
    saver.save(sess,