Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!newsfeed.internetmci.com!iol!tank.news.pipex.net!pipex!warwick!yama.mcc.ac.uk!thor.cf.ac.uk!news
From: Kyle Miller <millerka@cf.ac.uk>
Subject: Recurrent BP
Sender: news@cf.ac.uk (USENET News System)
Message-ID: <DnJHLB.9B7@cf.ac.uk>
Date: Thu, 29 Feb 1996 13:31:59 GMT
X-Nntp-Posting-Host: paul-2.engi.cf.ac.uk
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0
X-Mailer: Mozilla 1.22 (Windows; I; 32bit)
Organization: Cardiff University of Wales
Lines: 10

Has anyone here had problems with the backpropagtion algorithm 
when used with recurrent networks (in this case the Elman 
network), where the weights sometimes go off to infinity?

Or am I doing something wrong?

Please, please, tell me if you know the answer.

Email me!

