How to implement validation check using trainbr function in MATLAB 2010b?

2 views (last 30 days)
In previous MATLAB versions, until 2008b, it was possible to implement validation check and stop during NN training using trainbr function. From 2009 and 2010b version validation check can not be implemented is such function. Is there another NN training function in these versions which validation check can be implemented?

Accepted Answer

Greg Heath
Greg Heath on 20 Oct 2012
Probably all of them. However, I recommend
TRAINLM for regression/curvefitting with TRAINRP as a huge dataset alternative.
TRAINSCG for classification/patternrecognition.
Validation stopping is not necessary when using other techniques for mitigating overtraining an overfit nets. See the comp.ai.neural-nets FAQ.
Nevertheless, I would have liked to keep the option for combining the two.
Hope this helps.
Thank you officially accepting my answer.
Greg

More Answers (0)

Community Treasure Hunt

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

Start Hunting!