Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!news.ultranet.com!news.sprintlink.net!howland.reston.ans.net!news.moneng.mei.com!news.ecn.bgu.edu!siemens!flake
From: flake@scr.siemens.com (Gary William Flake)
Subject: Re: Gaussian transfer function ???
Message-ID: <D6M93s.1ML@scr.siemens.com>
Sender: news@scr.siemens.com (NeTnEwS)
Nntp-Posting-Host: gull.scr.siemens.com
Organization: Siemens Corporate Research, Princeton NJ
References: <1995Mar30.113145.1831@cathy.ijs.si> <3lrs37$69c@cantaloupe.srv.cs.cmu.edu> <mike.797036733@motion> <3m0bh8$g1@ninurta.fer.uni-lj.si>
Date: Thu, 6 Apr 1995 13:50:15 GMT
Lines: 28

Simon Dobrisek <simond@kette.fer.uni-lj.si> wrote:
> 
> Only a question:
> 
> Does exist 'a kind' of RBF 'neural network', which uses not 
> only the centroid vectors as the weight vectors and the 
> Euclidian distance (+ activation function) as the decision
> rule, but uses centroid vectors and covariance matrices as
> 'a kind' of the weight vectors and the Mahalanobis distance 
> (+activation function) as the decision rule?
> 
> Thanx!
> 
> Simon.

Take a look at Poggio and Girosi's work.

@TECHREPORT{GIR93,
        AUTHOR = {F.~Girosi and M.~Jones and T.~Poggio},
        TITLE = {Priors, Stabilizers and Basis Functions: from
                 regularization to radial, tensor and additiv splines},
        INSTITUTION = {MIT AI Laboratory},
        YEAR = {1993},
        NUMBER = {A.I. Memo No. 1430, C.B.C.L. Paper No. 75}
}
-- 
Gary W. Flake,  flake@scr.siemens.com,  Phone: 609-734-3676,  Fax: 609-734-6565
Siemens Corporate Research,  755 College Road East,  Princeton, NJ  08540,  USA
