Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!news.mathworks.com!nntp.primenet.com!cs.utexas.edu!newshost.convex.com!newsgate.duke.edu!interpath!news.interpath.net!sas!newshost.unx.sas.com!saswss
From: saswss@hotellng.unx.sas.com (Warren Sarle)
Subject: Re: Comparing trained NN models with other modeling techniques
Originator: saswss@hotellng.unx.sas.com
Sender: news@unx.sas.com (Noter of Newsworthy Events)
Message-ID: <Dy08ou.Jqu@unx.sas.com>
Date: Fri, 20 Sep 1996 00:14:54 GMT
X-Nntp-Posting-Host: hotellng.unx.sas.com
References:  <12SEP96.19281896@enh.nist.gov>
Organization: SAS Institute Inc.
Lines: 19


In article <12SEP96.19281896@enh.nist.gov>, ling@enh.nist.gov writes:
|> Can anyone comment on techniques to compare trained NN models with other
|> trained models, for example multiple linear regression.  Somebody
|> recommended one possibility, a t-test.  References and/or implementation
|> hints are greatly appreciated.

See Prechelt, L. (1994), "PROBEN1--A set of neural network benchmark
problems and benchmarking rules," Technical Report 21/94, Universitat
Karlsruhe, 76128 Karlsruhe, Germany,
ftp://ftp.ira.uka.de/pub/papers/techreports/1994/1994-21.ps.gz,
sections 2.8 and 3.3. For more information on t-tests, see any
introductory statistics textbook.

-- 

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.
