tensorflow学习：预测样本的分布（一）

程序功能：产生一百个样本点，样本点服从y = 0.3 * x + b，基于tensorflow利用梯度下降法学习w，b

import tensorflow as tf
import numpy as np

#产生一百个随机数
x_data = np.random.rand(100).astype(np.float32)
y_data = x_data * 0.3 + 0.1

#产生一个随机变量，范围在-10和+10之间
Weights = tf.Variable(tf.random_uniform([1], -10.0, 10.0))
biases = tf.Variable(tf.zeros([1]))

y = Weights * x_data + biases
#reduce_mean：取平均值
loss = tf.reduce_mean(tf.square(y - y_data))
#使用梯度下降法，并且学习步长是0.5
optimizer = tf.train.GradientDescentOptimizer(0.5)
train = optimizer.minimize(loss)

#这句话不能少，初始化所有变量
init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    
    for step in range(2000):
        sess.run(train)
        if step % 20 == 0:
            print (step, sess.run(Weights), sess.run(biases))