tensorflow——用RNN实现MNIST手写数字识别

import tensorflow as tf
import numpy as np

from tensorflow.examples.tutorials.mnist import input_data
from tensorflow.contrib import rnn

mnist=input_data.read_data_sets('./MNIST',one_hot=True)


chunk_size=28
chunk_n=28

rnn_size=256

n_output_layer=10

X=tf.placeholder('float',[None,chunk_n,chunk_size])
Y=tf.placeholder('float')
batch_size=100
def recurrent_neural_network(data):
    layer={'w_':tf.Variable(tf.random_normal([rnn_size,n_output_layer])),'b_':tf.Variable(n_output_layer)}
    data=tf.transpose(data,[1,0,2])
    data=tf.reshape(data,[-1,chunk_size])
    data=tf.split(data,chunk_n,0)
    '''
    with tf.variable_scope('cell_def',reuse=None):
        lstm_cell=rnn.BasicLSTMCell(rnn_size)
        #status=lstm_cell.zero_state(data,dtype=tf.float32)
    with tf.variable_scope('lstm_def',reuse=True):
        outputs,status=rnn.static_rnn(lstm_cell,data,dtype=tf.float32)
    '''
    lstm_cell = rnn.BasicLSTMCell(rnn_size)
    outputs, status = rnn.static_rnn(lstm_cell, data, dtype=tf.float32)
    
    output=tf.add(tf.cast(tf.matmul(outputs[-1], layer['w_']),tf.float32), tf.cast(layer['b_'],tf.float32))
    #output=tf.add(tf.matmul(outputs[-1], layer['w_']),tf.float32, layer['b_'],tf.float32)
    return output

def train_neural_network(X,Y):
    predict=recurrent_neural_network(X)
    cost_func=tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=predict,labels=Y))
    optimizer=tf.train.AdamOptimizer().minimize(cost_func)
    
    epochs=13
    with tf.Session() as session:
        session.run(tf.initialize_all_variables())
        epoch_loss=0
        for epoch in range(epochs):
            for i in range(int(mnist.train.num_examples/batch_size)):
                x,y=mnist.train.next_batch(batch_size)
                x=x.reshape([batch_size,chunk_n,chunk_size])
                _,c=session.run([optimizer,cost_func],feed_dict={X:x,Y:y})
                epoch_loss+=c
            print(epoch,':',epoch_loss)
            
        correct=tf.equal(tf.argmax(predict,1),tf.argmax(Y,1))
        accuracy=tf.reduce_mean(tf.cast(correct,'float'))
        print('准确率:',accuracy.eval({X:mnist.test.images.reshape(-1,chunk_n,chunk_size),Y:mnist.test.labels}))
        
train_neural_network(X,Y)