Prev: Macro excel
Next: fminunc optimization
From: Missam None on 10 Apr 2010 13:18
Hi

Each time,I train my network ,the training stops becaue "Minimum Gradient reached"
knowing that :inputs=4*4000
targets=14*4000,
Is it be possible to answer these questions:

1-Does this have anything to do with the inputs or targets size?
2-Does it have a relation with the way of initialising the network?if yes,please show me the right way to do it.

Thank you in advance.
Here is the code:

numHiddenNeurons = 7;
net = newpr(inputs,targets,numHiddenNeurons,{},'trainscg');
net.divideParam.trainRatio =60/100;
net.divideParam.valRatio = 20/100;
net.divideParam.testRatio =20/100;
net.trainParam.goal=0.01;
net.trainParam.mem_reduc=2;

net.initFcn='initlay';
net.layers{1}.initFcn='initwb';
net.layers{2}.initFcn='initwb';
net.inputWeights{1,1}.initFcn='rands';
net.layerWeights{1,2}.initFcn='rands';
net.biases{1}.initFcn='rands';
net.biases{2}.initFcn='rands';
net = init(net);

net.trainParam.min_grad=1e-30;
net.layers{1}.transferFcn='tansig';
net.layers{2}.transferFcn='hardlims';


% Train and Apply Network

[net,tr] = train(net,inputs,targets);
From: Missam None on 10 Apr 2010 17:17
Could anyone help??
 | 
Pages: 1
Prev: Macro excel
Next: fminunc optimization