import numpy as np
import matplotlib.pylab as plt
from pylab import mpl
mpl.rcParams['font.sans-serif'] = ['SimHei'] 

生成直线数据并加入噪音画图显示

# print(train_x)
noise = np.random.randn(*train_x.shape) * 0.3
train_y = 2 * train_x + noise  # 给每一个点加上噪音
# print(noise)
plt.plot(train_x, train_y, "go", label="我的初始数据")
plt.legend()
plt.show()

定义模型的输入和输出

y = tf.placeholder("float")
# 定义并初始化模型的权重偏置
w = tf.Variable(tf.random_normal([1]), name="weight")
b = tf.Variable(tf.zeros([1]), name="bias")
# 定义模型的前向传播过程
y_predict = tf.multiply(w, x) + b

定义模型的反向传播

定义损失函数

learning_rate = 0.01  # 定义学习率
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)  # 定义优化器

# 定义超参数
train_epochs = 100
display_epoch = 2

训练模型

    sess.run(init)
    plotdata = {"epochs": [], "cost": []}  # 保存训练得到的参数
    for epoch in range(train_epochs):
        for X, Y in zip(train_x, train_y):
            sess.run(optimizer, feed_dict={x: X, y: Y})

        if epoch % display_epoch == 0:
            print("训练的epoch为:%d cost为:%f %f" % (epoch+1, sess.run(cost, feed_dict={x: train_x, y: train_y}), sess.run(b)))
            plotdata["epochs"].append(epoch+1)
            plotdata["cost"].append(sess.run(cost, feed_dict={x: train_x, y: train_y}))

    print("训练完成")
    print("模型训练的结果为： ", "w", sess.run(w), "b:", sess.run(b), "cost:",
          sess.run(cost, feed_dict={x: train_x, y: train_y}))

    plt.plot(train_x, train_y, "go", label="我的初始数据")
    plt.plot(train_x, sess.run(w) * train_x + sess.run(b), label='Fitted line')
    plt.legend()
    plt.show()```