Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newshost.marcam.com!uunet!in1.uu.net!ulowell.uml.edu!umassd.edu!umassd!s_vogt
From: s_vogt@umassd.cis.umassd.edu (Stefan Vogt)
Subject: [Q] V&V techniques for ANN?
Message-ID: <D4Grrx.F5A@umassd.edu>
Sender: usenet@umassd.edu (USENET News System)
Organization: University of Massachusetts Dartmouth
X-Newsreader: TIN [version 1.2 PL2]
Date: Thu, 23 Feb 1995 17:40:43 GMT
Lines: 14


Hi,

I am woundering weather there are some techniques to Verify and
Validate the function of an ANN.
As long as we try test patterns we may observe 
    * some patterns not learned by the net;
    * an error over all test patterns.
But how can we analyse the output of the net in general? Maybe some
input patterns produce chatastrophic output values?

Any suggestion or references to that problem?

Bye, Stefan
