HBU深度学习作业4-BP算法

numpy

import numpy as np
import matplotlib.pyplot as plt

def sigmoid(x):
    return 1/(1+np.exp(-x))

#前向传播
def forward_propagation(x1,x2,w1,w2,w3,w4,w5,w6,w7,w8,y1,y2):
    h1=sigmoid(w1*x1+w3*x2)
    h2=sigmoid(w2*x1+w4*x2)
    o1=sigmoid(w5*h1+w7*h2)
    o2=sigmoid(w6*h1+w8*h2)
    print('前向传播：h1,h2,o1,o2:',round(h1,2),round(h2,2),round(o1,2),round(o2,2))
    error=((o1-y1)**2+(o2-y2)**2)/2
    print('均方误差error：',round(error,2))
    return h1,h2,o1,o2,error

#反向传播
#求梯度
def back_propagation(o1,o2,y1,y2,h1,h2,x1,x2,w1,w2,w3,w4,w5,w6,w7,w8):
    d_w8=(o2-y2)*o2*(1-o2)*h2
    d_w7=(o1-y1)*o1*(1-o1)*h2
    d_w6=(o2-y2)*o2*(1-o2)*h1
    d_w5=(o1-y1)*o1*(1-o1)*h1
    d_w4=(o1-y1)*o1*(1-o1)*w7*h2*(1-h2)*x2+(o2-y2)*o2*(1-o2)*w8*h2*(1-h2)*x2
    d_w3=(o1-y1)*o1*(1-o1)*w5*h1*(1-h1)*x2+(o2-y2)*o2*(1-o2)*w6*h1*(1-h1)*x2
    d_w2=(o1-y1)*o1*(1-o1)*w7*h2*(1-h2)*x1+(o2-y2)*o2*(1-o2)*w8*h2*(1-h2)*x1
    d_w1=(o1-y1)*o1*(1-o1)*w5*h1*(1-h1)*x1+(o2-y2)*o2*(1-o2)*w6*h1*(1-h1)*x1
    print('各权值的更新量：',round(d_w1,2),round(d_w2,2),round(d_w3,2),round(d_w4,2),round(d_w5,2),round(d_w6,2),round(d_w7,2),round(d_w8,2))
    return d_w1, d_w2, d_w3, d_w4, d_w5, d_w6, d_w7, d_w8

#更新权值
def updata_weight(w1,w2,w3,w4,w5,w6,w7,w8,d_w1,d_w2,d_w3,d_w4,d_w5,d_w6,d_w7,d_w8,step=1):
    w1=w1-step*d_w1
    w2=w2-step*d_w2
    w3=w3-step*d_w3
    w4=w4-step*d_w4
    w5=w5-step*d_w5
    w6=w6-step*d_w6
    w7=w7-step*d_w7
    w8=w8-step*d_w8
    print('更新后的权值：',round(w1,2),round(w2,2),round(w3,2),round(w4,2),round(w5,2),round(w6,2),round(w7,2),round(w8,2))
    return w1, w2, w3, w4, w5, w6, w7, w8

x1,x2=0.5,0.3
w1,w2,w3,w4,w5,w6,w7,w8=0.2,-0.4,0.5,0.6,0.1,-0.5,-0.3,0.8
y1,y2=0.23,-0.07
Error=[]
epoh=500#训练次数
step=eval(input('请输入步长：'))
for i in range(epoh):
    print('\n第{}轮:'.format(i+1))
    h1,h2,o1,o2,error=forward_propagation(x1,x2,w1,w2,w3,w4,w5,w6,w7,w8,y1,y2)#前向传播求出损失
    Error.append(error)
    d_w1,d_w2,d_w3,d_w4,d_w5,d_w6,d_w7,d_w8=back_propagation(o1, o2, y1, y2, h1, h2, x1, x2, w1, w2, w3, w4, w5, w6, w7,