Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!nntp.sei.cmu.edu!news.cis.ohio-state.edu!news.maxwell.syr.edu!cpk-news-hub1.bbnplanet.com!cam-news-hub1.bbnplanet.com!news.bbnplanet.com!news.mathworks.com!newsgate.duke.edu!interpath!news.interpath.net!news.interpath.net!sas!newshost.unx.sas.com!saswss
From: saswss@hotellng.unx.sas.com (Warren Sarle)
Subject: Re: RBF networks
Originator: saswss@hotellng.unx.sas.com
Sender: news@unx.sas.com (Noter of Newsworthy Events)
Message-ID: <E6By2K.AyM@unx.sas.com>
Date: Fri, 28 Feb 1997 20:25:32 GMT
X-Nntp-Posting-Host: hotellng.unx.sas.com
References:  <5ev58f$n2c@scotsman.ed.ac.uk>
Organization: SAS Institute Inc.
Lines: 28


In article <5ev58f$n2c@scotsman.ed.ac.uk>, pes@ee.ed.ac.uk (Paul E Strauch) writes:
|> 
|> In article <330DBC66.4A46@aprky.herts.ac.uk>, Dr Zhen-ni Wang <z.wang@aprky.herts.ac.uk> writes:
|> >I am looking at the fundamental difference between Radial basis function
|> >networks and normalized radial basis function networks* in terms of the
|> >rate of convergence and generalization capability.
|> >I wonder if anyone could help me with some advice or literature source.
... 
|> "Side effects of normalising radial basis function networks" R. Shorten,
|> R. Murray-Smith, International Journal of Neural Systems, Vol. 7, No. 2,
|> May 1996, p. 167-179.

Thanks for posting that reference. There are distressingly few
publications that acknowledge the very important differences between
ordinary RBF networks and normalised RBF networks. However, I think
that Shorten and Murray-Smith are excessively pessimistic regarding
NRBF networks. The problems they cite are mostly due to using unequal
widths. In most respects, NRBF networks with equal widths behave more
nicely than either ORBFs or MLPs. Details are in the FAQ under "How do 
MLPs compare with RBFs?" at ftp://ftp.sas.com/pub/neural/FAQ2.html .
-- 

Warren S. Sarle       SAS Institute Inc.   The opinions expressed here
saswss@unx.sas.com    SAS Campus Drive     are mine and not necessarily
(919) 677-8000        Cary, NC 27513, USA  those of SAS Institute.
 *** Do not send me unsolicited commercial or political email! ***

