Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!lerc.nasa.gov!magnus.acs.ohio-state.edu!freenet.columbus.oh.us!pacific.mps.ohio-state.edu!math.ohio-state.edu!jussieu.fr!oleane!plug.news.pipex.net!pipex!tube.news.pipex.net!pipex!tank.news.pipex.net!pipex!news.mathworks.com!news.kei.com!wang!news
From: bruck@actcom.co.il (Uri Bruck)
Subject: Re: Slow Learning or Rapid Flexibility? (was Re: Battlefield AI)
Reply-To: bruck@actcom.co.il
Organization: ACTCOM - Internet Services in Israel
Date: Wed, 8 Nov 1995 19:16:13 GMT
Message-ID: <DHqoBD.11E@actcom.co.il>
References: <446muf$i0c@nyx.cs.du.edu> <h3Hn4Tp.predictor@delphi.com> <46lebq$ap6@ixnews7.ix.netcom.com> <46mk3g$qgr@hermes.cair.du.edu> <47lapk$rj@dove.eecs.umich.edu>
Sender: news@wang.com
Lines: 20

rjones@dove.eecs.umich.edu (Randolph M. Jones) wrote:


>Thus, I think the kind of reactive, flexible opponents we want will arise
>(at least initially) not so much from giving the system a learning ability,
>but from building into the opponent a thorough knowledge of the game (by no
>means trivial) and the ability to map observed behavior onto internal game
>goals and strategies.  Certainly there would be room for the system to
>learn when it could not identify particular behaviors, but most of the
>rapid flexibility would not arise from learning.

>Comments?

>Randy

I think that what you're describing here is what everyone's been
talking about. It seems to me like you're discussing semantics.
Uri


