贝叶斯公式在全连接神经网络权重的应用

import numpy as np
import math

#三行为三组数据    
x = np.array([[1,  4,  3,  3,  2,  2,  3],
              [85, 76, 89, 64, 55, 98, 73],
              [2,  2,  3,  3,  1,  2,  3]])
                                                #x2 = np.array([100, 76, 89, 64, 55, 98, 73])
                                                #x3 = np.array([2, 2, 3, 3, 1, 2, 3])
t = np.array([])
y = np.array([]) 

sg = np.zeros(12)
w = np.ones(100)
w = w.reshape(10,10)

alp = 0.01
eps = 10e-5
e0 = e1 = e2 = 3

n = np.zeros(10)
b = np.zeros(10)

#求解 n4-n7 ：
for i in range(4, 8):
    for j in range(1, 4):
        n[i] += x[j-1, 0] * w[i, j] 
        
#加上启动函数后求解：n8 -n9
for i in range (4,8):
    for j in range(1,3):
        b[i] = (1/(1+math.exp(x[i-1,j-1])))*n[i]
        
for i in range(8,10):
    for j in range(4,9):
        n[i] += w[i,j]*b[j]
        
for i in range (20):
    
    
    sg1 = (y[i]-t[i])*(1/(1+math.exp(x[i])))*(1-((1/(1+math.exp(x[i])))))*n4

print(n[i])


    

import numpy as np
n = np.ones(10)
w = np.ones(100)
w = w.reshape(10,10)

x = np.array([[1, 4, 3, 3, 2, 2, 3],
              [85, 76, 89, 64, 55, 98, 73],
              [2, 2, 3, 3, 1, 2, 3]])
for i in range(4,8):
    for j in range (1,3):
        
        n[i] += w[i,j]*x[j,0]
    print(n[i])
        

88.0
88.0
88.0
88.0

import math
a = 10e-4
a

0.001