Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!satisfied.apocalypse.org!news.mathworks.com!newshost.marcam.com!uunet!psinntp!cronkite!blw
From: blw@utrc.utc.com (Brad Whitehall)
Subject: PAC ANN Papers
Message-ID: <1995Mar17.160952.15300@cronkite.res.utc.com>
Sender: news@cronkite.res.utc.com
Nntp-Posting-Host: 192.249.49.90
Organization: United Technologies Research Center
X-Newsreader: TIN [version 1.1 PL8]
Date: Fri, 17 Mar 1995 16:09:52 GMT
Lines: 22

 I am looking for some references to papers that have PAC
(or PAC like) results for multi-layer networks.
Also, I am looking for papers that provide PAC results
that deal with ANNs that are used to predict real
numbers, not just classes (0 or 1).  Can anyone point
me to some papers in this area?

I have read the papers:
"What size net gives valid generalization? by Baum and Haussler

and

Can neural networks do better than the Vapnik-Chervonenkis bounds? by
Tesauro and Cohn

These are good, but I am looking for the next step beyond what they
cover.

Thanks,
Brad Whitehall
blw@antares.res.utc.com

