发表于 | Edited on | 分类于 | 评论数：

Tensorflow学习3：非线性回归 

1
2
3
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
# 用numpy生成200个随机点
x_data = np.linspace(-0.5, 0.5, 200)[:,np.newaxis] #增加一个维度，变成单列矩阵
noise = np.random.normal(0,0.02,x_data.shape)
y_data = np.square(x_data) + noise

#
x = tf.placeholder(tf.float32,[None,1])
y = tf.placeholder(tf.float32,[None,1])

Weights_L1 = tf.Variable(tf.random_normal((1,10)))
biases_L1 = tf.Variable(tf.zeros((1,10)))
Wx_plus_b_L1 = tf.matmul(x, Weights_L1) + biases_L1
L1 = tf.nn.softplus(Wx_plus_b_L1)

Weights_L2 = tf.Variable(tf.random_normal((10,1)))
biases_L2 = tf.Variable(tf.zeros((1,1)))
Wx_plus_Biases_L2 = tf.matmul(L1, Weights_L2) + biases_L2
#predict_y = tf.nn.tanh(Wx_plus_Biases_L2)
predict_y = Wx_plus_Biases_L2

loss = tf.reduce_mean(tf.square(predict_y - y))
train = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for _ in range(1000):
        sess.run(train, feed_dict={x:x_data, y:y_data})
        
    _predict_y = sess.run(predict_y, feed_dict={x:x_data})
    plt.figure()
    plt.scatter(x_data, y_data)
    plt.plot(x_data, _predict_y, "red", lw=5)
    plt.show()

png

꧁༺The༒End༻꧂