TensorFlow之神经网络练习

TensorFlow实现神经网络：
           

import tensorflow as tf
from numpy.random import RandomState

batch_size=8
           

#定义权值、x、y
w1=tf.Variable(tf.random_normal([2,3],stddev=1,seed=1))
w2=tf.Variable(tf.random_normal([3,1],stddev=1,seed=1))
biases1=tf.Variable(tf.constant(0.1,shape=[3]))
biases2=tf.Variable(tf.constant(0.1,shape=[1]))
x=tf.placeholder(tf.float32,shape=(None,2),name="x-input")
y_=tf.placeholder(tf.float32,shape=(None,1),name="y-input")

           

#输入、中间层、输出计算
a=tf.nn.relu(tf.matmul(x,w1)+biases1)
y=tf.nn.relu(tf.matmul(a,w2)+biases2)
           

#定义损失函数、训练优化
cross_entropy=-tf.reduce_mean(y_*tf.log(tf.clip_by_value(y,1e-10,1.0)))
train_step = tf.train.AdamOptimizer(0.0001).minimize(cross_entropy)

           

#随机定义输入值
rdm = RandomState(1)
X = rdm.rand(128,2)
Y = [[int(x1+x2<1)] for (x1,x2) in X]
           

#生成会话执行
           

with tf.Session() as sess:
    init_op = tf.global_variables_initializer()
    sess.run(init_op)
    
    print "w1", sess.run(w1)
    print "w2", sess.run(w2)
    print "\n"
    #迭代
    STEPS = 10001
    for i in range(STEPS):
        start = (i*batch_size)%128
        end = start+batch_size
        sess.run(train_step,feed_dict={x:X[start:end],y_:Y[start:end]})
        if i % 1000 == 0:
            total_cross_entropy = sess.run(cross_entropy,feed_dict={x:X,y_:Y})
            print "After %d training step(s),cross entropy on all data is %g" % (i,total_cross_entropy)#输出损失值
            
    print "\n"
    print "w1:",sess.run(w1)
    print "w2:",sess.run(w2)