import tensorflow as tf

import numpy as np

#添加神經(jīng)層

def add_layer(inputs,in_size,out_size,activation_function=None):

? ? Weights = tf.Variable(tf.random_normal([in_size,out_size]))

? ? biases = tf.Variable(tf.zeros([1,out_size]) + 0.1)

? ? Wx_plus_b = tf.matmul(inputs,Weights) + biases

? ? if activation_function is None:

? ? ? ? outputs = Wx_plus_b

? ? else:

? ? ? ? outputs = activation_function(Wx_plus_b)

? ? return outputs

#設置輸入數(shù)據(jù)

x_data = np.linspace(-1,1,300)[:,np.newaxis]

#設置噪音數(shù)據(jù)

noise = np.random.normal(0,0.05,x_data.shape)

#設置預期輸出

y_data = np.square(x_data)-0.5 + noise

#設置傳入變量

xs = tf.placeholder(tf.float32,[None,1])

ys = tf.placeholder(tf.float32,[None,1])

#第一層，隱藏層，1個輸入，10個輸出（10個神經(jīng)元）

l1 = add_layer(xs,1,10,activation_function=tf.nn.relu)

#輸出層，一個輸出

prediction = add_layer(l1,10,1,activation_function=None)

#誤差/代價

loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),reduction_indices=[1]))

#最優(yōu)化過程

train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

init = tf.initialize_all_variables()

sess = tf.Session()

sess.run(init)

訓練和輸出

for i in range(1000):

? ? sess.run(train_step,feed_dict={xs:x_data,ys:y_data})

? ? if i % 50 == 0:

? ? ? ? print sess.run(loss,feed_dict={xs:x_data,ys:y_data})