Tensorflow构建神经网络（RNN , Mnist数据集）

二. 循环神经网络（RNN）

1.RNN神经网络的概念及相关介绍

• 概念：

       对时间序列上的变化进行建模的一种神经网络。

• 优点：

       基于之前的运行结果或者时间点，进行当前的预测。

2.代码部分

import tensorflow as tf
#引入RNN
from tensorflow.contrib import rnn
from tensorflow.examples.tutorials.mnist input_data
mnist = input_data.read_data_sets("data/",one_hot=True)

#定义参数
batch_size = 128

#定义训练数据
x = tf.placeholder("float",[None,28,28])
y = tf.placeholder("float",[None,10])

#定义W和b
weights = {
    'out':tf.Variable(tf.random_normal([128,10]))
}
biases = {
    'out':tf.Variable(tf.random_normal([10]))
}

def RNN(x,weights,biases):
    #按照RNN的方式处理输入层
    x = tf.unstack(x,28,1)
    #lstm层
    #forget_bias (默认值为1)到遗忘门的偏置，为了减少在开始训练时遗忘的规模。
    lstm_cell = rnn.BasicLSTMCell(128,forget_bias=1.0)
    #获取lstm层的输出
    outputs,states = rnn.static_rnn(lstm_cell,x,dtype=tf.float32)
    #得到最后一层的输出
    return tf.matmul(outputs[-1],weights['out']) + biases['out']

#预测值
pred = RNN(x,weights,biases)

#定义代价函数和最优算法

#寻找全局最优点的优化算法，引入二次方梯度校正
#AdamOptimizer 不容易陷入局部最优点，速度更快
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred,labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=0.001).minimize(cost)

#结果对比
correct_pred = tf.equal(tf.argmax(pred,1),tf.argmax(y,1))

#求正确率
accuracy = tf.reduce_mean(tf.cast(correct_pred,tf.float32))

#初始化所有参数
init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    step = 1
    
    while step * batch_size < 100000:
        batch_x,batch_y = mnist.train.next_batch(batch_size)
        batch_x = batch_x.reshape((batch_size,28,28))
        sess.run(optimizer,feed_dict={x:batch_x,y:batch_y})
        if step % 10 == 0:
            acc = sess.run(accuracy,feed_dict={x:batch_x,y:batch_y})
            loss = sess.run(cost,feed_dict={x:batch_x,y:batch_y})
            print("Tter "+str(step * batch_size) + ",Minibatch Loss="+"{:.6f}".format(loss)+",Training Accuracy="+"{:.5f}".format(acc))
        step += 1
    print("Optimization Finished!")

    #数据验证
    test_len = 128
    test_data = mnist.test.images[:test_len].reshape((-1,28,28))
    test_label = mnist.test.labels[:test_len]
    print("Testing Accuracy:",sess.run(accuracy,feed_dict={x:test_data,y:test_label}))