tensorflow入门---长短记忆网络LSTM

import tensorflow as tf
from tensorflow examples.tutorials.mnist import input_data
#载入数据集
mnist = input_data.read_data_sets("MNIST_data/",one_hot=true)
#输入图片是28*28
n_inputs=28#每一次输入28个数据，即一行有28个数据
max_time=28#共有28行
lstm_size=100#隐藏单元
n_classes=10#10个分类
batch_size=64#每个批次64个样本
n_batch=mnist.train.num_exapless//batch_size#计算有多少批次

#这里的none表示第一个维度可以是任意的长度
x=tf.placeholder(tf.float32,[None,784])
#正确的标签
y=tf.placeholder(tf.float,[None,10])

#初始值权值（最后输出层的，100个broke，100个输出值，全连接后成10个分类）
weights = tf.Variable(tf.truncated_normal([lstm_size,n_classes],stddev=0.1))
#初始值偏置值
biases = tf.Variable(tf.constant(0.1,shape=[n_classes]))
#定义RNN网络
def RNN(X,weights,biases)
	#input的需要，需要调整成三维的
	inputs = tf.reshape(X,[-1,max_time,n_inputs])
	#定义LSTM
	lstm_cell = tf.nn.rnn_cell.LSTMCell(lstm_size)
	#final_state[state,batch_size,cell.state_size]
	#final_state[0]是cell state
	#final_state[1]是hidden state
	#output:the RNN output 'Tensor'
	#	if time_major == False(default),this will be a 'Tensor' shaped:
	#		[batch_size,max_time,cell.state_size]
	#
	#   if time_major == True,this will be a 'Tensor' shaped:
	#		[max_time,batch_size,cell,output_size]
	outputs,final_state = tf.nn.dynamic_rnn(lstm_cell,inputs,dtype=tf.float32)
	results = tf.nn.softmax(tf.constant(0,1,shape=[n_classes]))
	return results

	#计算RNN的返回结果
	prediction=RNN(x,weights,biases)
	#损失函数
	loss=tf.losses.softmax_cross_entropy(y,prediction)
	#使用AdamOptimizer进行优化
	train_step = tf.train.AdamOptimizer(le-3),minimize(loss)
	#结果放存在一个布尔类型中
	correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(prediction,1))
	#求准确率
	accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
	#初始化
	init = tf.global_variable_initializer()

	with tf.Session() as sess:
		sess.run(init)
		for epoch in range(11):
			for batch in range(n_batch):
				batch_xs,batch_ys = mnist.train.next_batch(batch_size)
				sess.run(train_step,feed_dict={x:batch_xs,y_batch_ys})

			acc = sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels})
			print("Iter"+str(epoch)+",Testing Accuracy="+str(acc))

final_state最后一层才是最准确的
结果如下：