:Ŵ㷨߼(Logical Function)Sigmoid

ʽ:

[a,er,l]=SigmoidFitGA(x,y,a0,MaxA,MinA,LoopSum,Error,Nbird)
[a,er,l]=SigmoidFitGA(x,y,a0,MaxA,MinA,LoopSum,Error)
[a,er,l]=SigmoidFitGA(x,y,a0,MaxA,MinA,LoopSum)
[a,er,l]=SigmoidFitGA(x,y,a0,MaxA,MinA)
[a,er,l]=SigmoidFitGA(x,y,a0,MaxA)
[a,er,l]=SigmoidFitGA(x,y,a0)
[a,er,l]=SigmoidFitGA(x,y)

x:һm*nݾ,ÿһжӦ
y:һm*1ľ,ÿһжӦxеÿһ
a0:ϵĳʼֵ,ΪΪ1*(n+1)ľ,ʾÿϱֵ,ĬΪ0
MaxA:Ӧa0бϵ,ĬΪ10
MinA:Ӧa0бϵ,ĬΪ-10
LoopSum:ѭ,ĬΪ3000
Error:ѭĿ,ĬΪ1E-16
Nbird:Ⱥ,ĬΪ200

a:صϵ
er:صĲвƽ
l:صյ

ԭ:

1ϵĺʽΪ,x(1)x(2)x(n)ֱxĵ12n
$$
   y = \dfrac{1}{1 + \exp(a_0 + \sum_{i=1}^n a_i x_i)}
$$
2ʽʼŴ㷨֮ǰȽз̱

3ȻŴ㷨

:

x={Rand<\rand>}(10,20,3);
x1=x(,1);
x2=x(,2);
x3=x(,3);
z=2+0.1*x1-0.3*x2+0.45*x3;
y=1/(1+{Exp<\Exp>}(z));

//ͨĴxyԭʼ
[a,b,l]=SigmoidFitGA(x,y)//ִбԵõµĽ
a =
[ 7.51216718132387    1.42291528655037    0.43257471980471   -2.89482372826135 ]
b =
[ 7.1268526753E-19 ]
l =
[ 11.0000000000000 ]