Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!swrinde!newsfeed.internetmci.com!info.ucla.edu!agate!sunsite.doc.ic.ac.uk!warwick!bsmail!usenet
From: M.J.Ratcliffe@bris.ac.uk
Subject: yet another plea for help re RBFs
Content-Type: text/plain; charset=us-ascii
Message-ID: <DpwwJ5.4vG@uns.bris.ac.uk>
Sender: usenet@uns.bris.ac.uk (Usenet news owner)
Nntp-Posting-Host: pc53.aer.bris.ac.uk
Content-Transfer-Encoding: 7bit
Organization: University of Bristol, UK
Mime-Version: 1.0
Date: Mon, 15 Apr 1996 16:30:41 GMT
X-Mailer: Mozilla 1.1N (Windows; I; 16bit)
Lines: 13

Does anybody understand the orthogonal least squares algorithm for 
selecting centres of an RBF network? I know that it selects centres based 
on whichever data point causes the maximum decrease in error. The obvious 
way to do this is to calculate the errors associated with each data 
point, and then perfom a full solution. This expensive technique is 
circumvented by the OLS algorithm in some mysterious way. I do not 
understand the bit about HOW it calculates the errors due to each data 
point.

Any help much appreciated 

Max.

