学会用tensorflow搭建简单的神经网络 2-1 用matplotlib画出变化图

本篇文的代码是针对于学会用tensorflow搭建简单的神经网络 2那篇文章的一个改进，重点在于让读者学会用matplotlib画出随机生成的一个测试数据，同时也画出了预测曲线，让程序员更加清楚地知道本训练中计算机学习的效果。  在阅读本文之前，建议先阅读《学会用tensorflow搭建简单的神经网络 2》这篇文章。

代码中红色部分为新添代码

#!/usr/bin/env python
# _*_ coding: utf-8 _*_
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
#add_laye
def add_layer(inputs, in_size, out_size, activation_function=None):
    # add one more layer and return the output of this layer
    Weights = tf.Variable(tf.random_normal([in_size, out_size]))
    biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
    Wx_plus_b = tf.matmul(inputs, Weights) + biases
    if activation_function is None:
        outputs = Wx_plus_b
    else:
        outputs = activation_function(Wx_plus_b)
    return outputs
# Make up some real data
x_data = np.linspace(-1,1,300)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape)
y_data = np.square(x_data) - 0.5 + noise
# define placeholder for inputs to network
xs = tf.placeholder(tf.float32, [None, 1])
ys = tf.placeholder(tf.float32, [None, 1])
# add hidden layer
layer1 = add_layer(xs, 1, 10, activation_function=tf.nn.relu)
# add output layer
prediction = add_layer(layer1, 10, 1, activation_function=None)
# the error between prediciton and real data
loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction),
                     reduction_indices=[1]))
#Select the optimizer to minimize the loss
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)
# Important step---Initializes all variables
init=tf.initialize_all_variables()
# Sess.run ---it will start the operation
sess=tf.Session()
sess.run(init)
#Iterations of 201 times
39 fig=plt.figure()  #生成一个图片框
40 ax=fig.add_subplot(1,1,1)#生成连续性的图片框
41 ax.scatter(x_data,y_data) #将训练数据以点的形式运行出来
42 plt.ion()  #连续生成图片
43 plt.show() #打印图片
for i in range(1000):
    sess.run(train_step, feed_dict={xs: x_data, ys: y_data})
    if i % 50 == 0:
        # to see the step improvement
        #print(sess.run(loss, feed_dict={xs: x_data, ys: y_data}))
49         try:
50             ax.lines.remove(lines[0])  # 因为线是在不断地生成，所以要把旧的线去掉，只显示新生成的，这样方便查看结果
51         except Exception:
52             pass
53         prediction_value=sess.run(prediction,feed_dict={xs: x_data, ys: y_data})  #run prediction
54         lines=ax.plot(x_data,prediction_value,'r-',lw=5)  #把prediction_value的数据plot上去，以红色的线，宽度为5
55         plt.pause(0.2) #每次新生成的线暂停0.2s

结果：截取其中一张图片