It is necessary to reset (initialize) the weights during the learning period??

I am working on forecasting of radiation solar by Matlab and I have just beginner’s expertise with the program. it is necessary to reset (initialize) the weights during the learning period to determine the number of hidden layer neurons ? How I can do it? What is the command? this is a part of my program :
please Where can I put the command to reset the weights ???!
n=50; % number of epochs
net=fitnet(NNHL) % NNHHL: number of nodes in hiden layer
for NNHL=1:15
net.layers{1}.dimensions=NNHL;
fprintf('number of nodes is :%d\n',NNHL);
for j=1:1:n % n: number of epochs
[net,tr] = train(net,inputs,targets);
outputs = net(inputsTesting);
perf = mse(net,targetsTesting,outputs);
end
end
best regards

Réponses (0)

Cette question est clôturée.

Tags

Clôturé :

le 20 Août 2021

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by