Parameters of Neural Network estimation methods in Matlab

Which method is usually used in matlab to estimate the net parameters? It is the maximum likelihood estimation or the Least Square method?

When using "trainbr" function to train the network (training with regularization) which method is used? It is true that the Least Square method can not be applied in this case because the beta and alpha parameters (alpha*sse+beta*W'W where W is the parameter vector) couldn't be found this way?

Answer 1

[Greg Heath]

Usually? The number of training algorithms is in double figures. Most likely, the one used most is FITNET for Regression and Curve-fitting. You can answer your own questions about the others as follows (Notice when to omit the ending semicolon)

 >> net          = fitnet;
    trainFcn   = net.trainFcn  
    performFcn = net.performFcn
    help trainlm

 >> net.trainFcn = 'trainbr';
    performFcn = net.performFcn

Hope this helps.

Thank you for formally accepting my answer

Greg