TensorFlow基础7——完整神经网络栗子

完整神经网络栗子：
数据加入噪音，更具真实性

 1 import tensorflow as tf 
 2 import numpy as np 
 3 
 4 def add_layer(input,in_size,out_size,activation_function=None):
 5     Weights = tf.Variable(tf.random_normal([in_size,out_size]))
 6     biases = tf.Variable(tf.zeros([1,out_size])+0.1)
 7     Wx_plus_b = tf.matmul(input,Weights) + biases
 8     if activation_function is None:
 9         outputs = Wx_plus_b
10     else:
11         outputs = activation_function(Wx_plus_b)
12     return outputs
13 
14 x_data = np.linspace(-1,1,300)[:,np.newaxis]
15 noise = np.random.normal(0,0.05,x_data.shape)#噪音
16 y_data = np.square(x_data)-0.5+noise
17 
18 xs = tf.placeholder(tf.float32,[None,1])
19 ys = tf.placeholder(tf.float32,[None,1])
20 
21 l1 = add_layer(xs,1,10,activation_function=tf.nn.relu)
22 predition = add_layer(l1,10,1,activation_function=None)
23 
24 loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys-predition),reduction_indices=[1]))
25 train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)
26 
27 init = tf.initialize_all_variables()
28 sess = tf.Session()
29 sess.run(init)
30 for i in range(1001):
31     sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
32     if i % 50 == 0:
33         print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))

结果：

　　损失函数越小，说明拟合越好（有可能过拟合）