Hey, On Mon, Dec 23, 2013 at 03:39:12PM -0200, Felipe Duque wrote: > Hello. I've used SOM in MVPA2 for some academic works before and have never > run into any kind of trouble. However, in my most recent work, I'm running > into overflow issues. My input data is [0, 1] normalized. > > Delving into som.py code, I'm afraid I didn't notice any kind of > 'forgetting' term in the training methods. Is this so? Why is this > saturation taking place?
I am not 100% sure what kind of issue you are seeing. Numerical overflow? The SOM code is pretty simplistic -- once we know what the issue is, it should be possible to determine its origin. Do you maybe have some demo data and code to reproduce the problem? Michael -- Michael Hanke http://mih.voxindeserto.de _______________________________________________ Pkg-ExpPsy-PyMVPA mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
