tensorflow使用基础（2）-- 非线性回归

非线性回归

import tensorflow as tf
import numpy as np

• 生成饱含随机噪音的数据

#生成饱含随机噪音的数据
x_data = np.linspace(-1.0, 1.0, 200)
noise = np.random.normal(0, 0.02, 200)
y_data = x_data ** 2 + noise
print(x_data[0:10],'\n',y_data[0:10])

[-1.         -0.98994975 -0.9798995  -0.96984925 -0.95979899 -0.94974874
 -0.93969849 -0.92964824 -0.91959799 -0.90954774] 
 [0.98826083 1.00405787 0.93467031 0.93196639 0.9507247  0.89131818
 0.8945671  0.90021127 0.83955296 0.82885712]

• 对x，y进行结构处理

#对x，y进行结构处理，变成列向量
x_data = np.reshape(x_data, (200, 1))
y_data = np.reshape(y_data, (200, 1))
print(x_data.shape,y_data.shape)

(200, 1) (200, 1)

- 创建三层的神经网络

神经网络的结构为 [1, 8, 1]
对于1,8之间，有权重 w.shape = (1,8), b.shape() = (1,8)
对于8,1之间，有权重 w.shape = (8,1), b.shape() = (1,1)

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

L1_weight = tf.Variable(tf.random_normal([1,8]))
L1_bias = tf.Variable(tf.random_normal([1,8]))
L1_z = tf.matmul(x, L1_weight) + L1_bias
L1_out = tf.nn.tanh(L1_z)

L2_weight = tf.Variable(tf.random_normal([8,1]))
L2_bias = tf.Variable(tf.random_normal([1,1]))
L2_z = tf.matmul(L1_out, L2_weight) + L2_bias
L2_out = tf.nn.tanh(L2_z)

loss = tf.reduce_mean(tf.square(y - L2_out))
train = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

init = tf.global_variables_initializer()

import matplotlib.pyplot as plt
with tf.Session() as sess:
    sess.run(init)
    for _ in range(1000):
        sess.run(train,feed_dict={x:x_data,y:y_data})
    prediction_value = sess.run(L2_out,feed_dict={x:x_data})
    plt.figure()
    plt.scatter(x_data,y_data)
    plt.plot(x_data,prediction_value,'r-')
    plt.show()