Page 43 - Algorithm Collections for Digital Signal Processing Applications using MATLAB

1. Artificial Intelligence 31

0.2225 1.0172
0.2426 1.0483
0.3244 1.1747

After rounding OUTPUT exactly matches with the Desired Output as shown
below
0 0
0 1
0 1
0 1
0 1
0 1
0 1
0 1

4.4 M-program for Training the Artificial Neural
Network for the Problem Proposed in the Previous
Section

___________________________________________________________
anngv.m
SSE=[ ]; INDEX=[ ];
INPUT=[0 0 0;0 0 1;0 1 0;0 1 1;1 0 0;1 0 1;1 1 0;1 1 1];
DOUTPUT=[0 0;0 1;0 1 ;0 1;0 1;0 1;0 1 ;1 1];
W1=rand(3,1);W2=rand(1,2);
B1=rand(1,1); B2=rand(1,2);
for i=1:1:1200
for j=1:1:8
H=logsiggv(INPUT*W1+repmat(B1,[8 1]));
OUTPUT=H*W2 +repmat(B2,[8 1]);
ERR_HO=DOUTPUT - OUTPUT;
ERR_IH=H.*(1-H).*(sum(ERR_HO')');
lr_ho=0.01;
lr_ih=0.01;
W2(1,1)=W2(1,1)+ERR_HO(j,1)*H(j,1)*lr_ho;
W2(1,2)=W2(1,2)+ERR_HO(j,2)*H(j,1)*lr_ho;
W1(1,1)=W1(1,1)+ERR_IH(j,1)*INPUT(j,1)*lr_ih;
W1(2,1)=W1(2,1)+ERR_IH(j,1)*INPUT(j,2)*lr_ih;
W1(3,1)=W1(3,1)+ERR_IH(j,1)*INPUT(j,3)*lr_ih;
B1=B1+ERR_IH(j,1)*lr_ih;
B2(1,1)=B2(1,1)+ERR_HO(j,1)*lr_ho;