Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!newsstand.cit.cornell.edu!portc01.blue.aol.com!news-peer.gsl.net!news.gsl.net!news.mathworks.com!uunet!in1.uu.net!news!usenet
From: "Larry" <adamsl@encorp.com>
Subject: Re: [Q] Preptogramming Hopfield-nets for solving TSP?
Message-ID: <01bbbb68$470651d0$c7962dc7@adamsl>
Sender: usenet@news.ezlink.com (usenet news)
Organization: EZLink Internet Access Fort Collins Colorado
Date: Wed, 16 Oct 1996 13:42:41 GMT
References: <3260146D.6591@alf.tel.hr>
X-Newsreader: Microsoft Internet News 4.70.1155
X-Nntp-Posting-Host: rsslip.ezlink.com
Lines: 24

Look up "simulated annealing". 

Marin Bonacci <marin.bonacci@alf.tel.hr> wrote in article
<3260146D.6591@alf.tel.hr>...
> Hi,
> 
> I'm a comp-sci student with some interest in ANNs. Since there are no
> resources about this subject in my school, this is the only place I
> could think of to ask this question. 
>   
> I read a few books about ANNs looking for a way to preprogramm Hopfield
> net to solve the Travelling Salesman Problem. None of the rules for
> setting the weights do not work! I tried to experiment, but could not
> come up with anything. It looks like the solution pattern IS a stable
> state, but it is not the only one. So the network just drops to one of
> the stable states (which by the way is never the right one), and stays
> there. 
> 
> I don't know if any of this makes any sense, but if someone knows
> anything about this, please answer.
> 
> 					Thanks, 
> 						Marin.
> 
