代码

import tensorflow as tf 
import numpy as np 
import matplotlib.pyplot as plt 
# generate data
data_size = 100
x_data = np.random.rand(data_size) * 0.6 + 0.2
noise = np.random.rand(data_size)
noise = (noise - noise.mean()) * 0.2
w = 1.5
b = 0.5
y_data = x_data * w + b + noise
train_test_boundary = int(0.7 * data_size)
x_train = x_data[:train_test_boundary]
y_train = y_data[:train_test_boundary]
x_test = x_data[train_test_boundary:]
y_test = y_data[train_test_boundary:]
# my model
train_loss_history = []
test_loss_history = []
nb_epoch = 100
my_w = 0
my_b = 0
learning_rate = 0.1
with tf.Graph().as_default():
    x = tf.placeholder(dtype = tf.float32, shape = [None])
    y_ = tf.placeholder(dtype = tf.float32, shape = [None])
    w = tf.Variable(0, dtype = tf.float32)
    b = tf.Variable(0, dtype = tf.float32)
    y = x * w + b
    loss = tf.reduce_mean((y - y_) ** 2)
    train = tf.train.AdamOptimizer(learning_rate).minimize(loss)
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for i in range(nb_epoch):
            sess.run(train, feed_dict = { x: x_train, y_: y_train})
            l = sess.run(loss, feed_dict = { x: x_train, y_:y_train})
            train_loss_history.append(l)
            l = sess.run(loss, feed_dict = { x: x_test, y_: y_test})
            test_loss_history.append(l)
        my_w = w.eval()
        my_b = b.eval()
# display
plt.figure(figsize=(8, 4))
plt.subplot(1, 2, 1)
plt.xlim([0, 1])
plt.ylim([0, 3])
plt.plot(np.linspace(0, 1, 100), my_w * np.linspace(0, 1, 100) + my_b, 'r--')
plt.scatter(x_train, y_train, 1.0, 'b', marker='o')
plt.title('dataset & hypothesis function')
plt.subplot(1, 2, 2)
plt.title('training & testing loss')
plt.plot(train_loss_history, 'b--')
plt.plot(test_loss_history, 'r--')
plt.legend(['trian', 'test'])
plt.show()