用于预测和误差分析的 BP 神经网络的 Python 实现（附源代码）
最编程
2024-03-02 16:15:53
...
import numpy as np
import matplotlib.pyplot as plt
from pylab import mpl
# 设置显示中文字体
import matplotlib; matplotlib.use('TkAgg')
mpl.rcParams["font.sans-serif"] = ["SimHei"]
matplotlib.rcParams['font.family'] = 'SimHei'
matplotlib.rcParams['font.size'] = 10
matplotlib.rcParams['axes.unicode_minus']=False
def sigmoid(x):
return 1/(1+np.exp(-x))
#前向计算
def forward_NN(x,w,b):
h1=sigmoid(w[0]*x[0]+w[1]*x[1]+b[0])
h2=sigmoid(w[2]*x[0]+w[3]*x[1]+b[0])
h3=sigmoid(w[4]*x[0]+w[5]*x[1]+b[0])
print(h1,h2,h3)
o1=sigmoid(w[6]*h1+w[8]*h2+w[10]*h3+b[1])
o2=sigmoid(w[7]*h1+w[9]*h2+w[11]*h3+b[1])
return h1,h2,h3,o1,o2
#反向传递 调整参数
def fit(o1,o2,y,x,w,lrate,epochs):
for i in range(epochs):
#循环迭代 调整参数
p1=lrate*(o1-y[0])*o1*(1-o1)
p2 = lrate * (o2 - y[1]) * o2 * (1 - o2)
w[0]=w[0]-(p1*w[6]+p2*w[7]*h1*(1-h1)*x[0])
w[1] = w[1] - (p1 * w[6] + p2 * w[7] * h1 * (1 - h1) * x[1])
w[2] = w[2] - (p1 * w[8] + p2 * w[9] * h2 * (1 - h2) * x[0])
w[3] = w[3] - (p1 * w[8] + p2 * w[9] * h2 * (1 - h1) * x[1])
w[4] = w[4] - (p1 * w[10] + p2 * w[11] * h3 * (1 - h3) * x[0])
w[5] = w[5] - (p1 * w[10] + p2 * w[11] * h3 * (1 - h3) * x[1])
w[6]=w[6]-p1*h1
w[7] = w[7] - p2 * h1
w[8] = w[8] - p1 * h2
w[9] = w[9] - p2 * h2
w[10] = w[10] - p1 * h3
w[11] = w[11] - p2 * h3
return w
print("步骤一 初始化参数")
x=[3,6]
y=[0,1]
w=[0.6,0.61,0.62,0.63,0.64,0.65,0.66,0.67,0.68,0.69,0.70,0.71,0.72,0.73]
b=[0.3,0.6]
lrate=0.4
print("步骤二 fit")
print("步骤三 预测")
print("真值为",y)
sumDs=[]
for epochs in range(0,101,5):
h1,h2,h3,o1,o2=forward_NN(x,w,b)
w=fit(o1,o2,y,x,w,lrate,epochs)
h1,h2,h3,o1,o2=forward_NN(x,w,b)
print("迭代",epochs,"次的输出为\n",o1,o2)
sumDs.append(o1-y[0]+(o2-y[1]))
print("画图")
plt.plot(range(0,101,5),sumDs)
plt.title("the epoch-error plot for 铅钡")
plt.xlabel("epochs")
plt.ylabel("totol error")
plt.show()