TrainInput = [0 0; 0 1; 1 0; 1 1]';
TrainTarget = [0; 1; 1; 0]';
net = newff(TrainInput, TrainTarget, 2, {'tansig' 'purelin'});
net.divideFcn = '';
net.trainParam.lr = 0.01;
net = init(net); %# init
net = train(net, TrainInput, TrainTarget);
Predict = net(TrainInput);
"This is my code in MATLAB"
Predict in MATLAB:
3.45168338355961e-12
0.999999999996918
1.00000000000490
1.32027722088424e-12
net.IW{1,1} =
-2.0053 -2.1062
-1.9156 -1.9577
net.LW{2,1} =
1.0728 -1.0657
net.b{1,1} =
2.1195
-1.6549
net.b =
[2x1 double]
[ -1.0320]
But I use same weight in Microsoft Excel for prove model.
Then Predict in Excel:
1.000892957
0.046416854
0.153997283
-0.99993138
Why aren't they equal?
