Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!newsfeed.internetmci.com!news.ee.net!news2.ee.net!ragnarok.oar.net!malgudi.oar.net!utnetw.utoledo.edu!news
From: dweddin@uoft02.utoledo.edu
Subject: Backprop LAMBDA question
Message-ID: <DnKx8o.JKE@utnetw.utoledo.edu>
Lines: 22
Sender: news@utnetw.utoledo.edu (News Manager)
Organization: University of Toledo
Date: Thu, 29 Feb 1996 22:07:58 GMT


Hey all,

for Backprop neural nets, what is a good rule of thumb
to set the LAMBDA value at.  should it be large to
approach a true step function or smaller to make a
gradule incline.

does it vary from application to application?

hey, i wanna know!   ;-)


thanks,

by the way, i've worked a lot with RBF's and would be happy
to answer questions on that topic since so many folks have
been so helpful with my backprop questions!


Don Wedding

