TensorFlow-线程回归模型

实验目的：

方程：y = Wx + b

通过大量的（x, y）坐标值，模型可以计算出接近W和b的值

实验步骤：

第一步：生成线程回归方程模型所需要的数据

import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt

# 随机生成1000个点，围绕在y=0.1x+0.3的直线周围
num_points = 1000
vectors_set = []
for i in range(num_points):
    x1 = np.random.normal(0.0, 0.50)  # 正态分布 0.0:均值，0.50：标准差
    y1 = x1 * 0.1 + 0.3 + np.random.normal(0.0, 0.03)
    vectors_set.append([x1, y1])

# 生成一些样本数据
x_data = [v[0] for v in vectors_set]
y_data = [v[1] for v in vectors_set]
# print(x_data) 会有1000个数据
plt.scatter(x_data, y_data, c='r')
plt.show()

第二步：建立线性回归模型，将生成的数据（x_data，y_data）喂给模型，并产生结果。

# 生成1维的W矩阵， 取值是[-1,1]之间的随机数
W = tf.Variable(tf.random_uniform([1], -1.0, 1.0), name='W')
# 生成1维的b矩阵，初始值是0
b = tf.Variable(tf.zeros([1]), name='b')
# 记过计算得出预估值y
y = W * x_data + b

# 以预估值y和实际值y_data之间的均方误差作为损失
loss = tf.reduce_mean(tf.square(y - y_data, name='loss'))
# 采用梯度下降法来优化参数
optimizer = tf.train.GradientDescentOptimizer(0.5)
# 训练的过程就是最小化这个误差值
train = optimizer.minimize(loss, name='train')

sess = tf.Session()
# 初始化sess
init = tf.global_variables_initializer()
sess.run(init)

# 初始化的W和b是多少
print("W = ", sess.run(W), "b = ", sess.run(b), "loss = ", sess.run(loss))
# 执行20次训练
for step in range(20):
    sess.run(train)
    # 输出寻来你好的W和b
    print("W = ", sess.run(W), "b = ", sess.run(b), "loss = ", sess.run(loss))

结果：

W =  [0.6756458] b =  [0.] loss =  0.16456214
W =  [0.53388023] b =  [0.2748111] loss =  0.050819542
W =  [0.4182969] b =  [0.28113642] loss =  0.027618099
W =  [0.33377516] b =  [0.28629357] loss =  0.015202045
W =  [0.2719439] b =  [0.2900648] loss =  0.0085575525
W =  [0.22671175] b =  [0.29282364] loss =  0.00500173
W =  [0.19362253] b =  [0.29484183] loss =  0.0030988192
W =  [0.16941638] b =  [0.2963182] loss =  0.0020804694
W =  [0.15170857] b =  [0.29739827] loss =  0.0015354961
W =  [0.13875458] b =  [0.29818836] loss =  0.0012438523
W =  [0.12927818] b =  [0.29876634] loss =  0.0010877779
W =  [0.12234581] b =  [0.29918915] loss =  0.0010042539
W =  [0.11727448] b =  [0.29949847] loss =  0.0009595558
W =  [0.11356459] b =  [0.29972476] loss =  0.0009356354
W =  [0.11085065] b =  [0.29989028] loss =  0.00092283427
W =  [0.10886529] b =  [0.30001137] loss =  0.0009159839
W =  [0.10741292] b =  [0.30009997] loss =  0.0009123176
W =  [0.10635044] b =  [0.30016476] loss =  0.00091035594
W =  [0.1055732] b =  [0.30021217] loss =  0.0009093059
W =  [0.10500462] b =  [0.30024683] loss =  0.00090874406
W =  [0.10458867] b =  [0.30027223] loss =  0.00090844336

我们可以看到W不断趋近于0.1，b不断趋近于0.3，loss不断变小。

说明模型是可用的。