Michael+Gherrity


 * Home * People * Michael Gherrity**

toc an American computer scientist and AI-researcher from the [|University of California, San Diego]. He made his Ph.D. in 1993, //A Game Learning Machine//, elaborating on SAL (Search and Learn), his General Game Playing program. While applying a move generator, and losing if own king is captured as sole domain specific chess knowledge, it was the first chess program used Temporal Difference Learning. In a match of 4200 games against GNU Chess (One second per move), it started to play random moves within its two ply search plus Consistency Search, a generalized Quiescence Search, but learned to play reasonable, but still weak chess. It archived eight draws, apparently due to a repetition detection bug in GNU Chess.
 * Michael (Mike) Gherrity**,

=Selected Publications=
 * Michael Gherrity (**1993**). //A Game Learning Machine//. Ph.D. thesis, [|University of California, San Diego], advisor Paul Kube,  [|pdf], [|pdf]
 * Michael Gherrity, Paul Kube (**1993**). //Quiescent Search is Beneficial.// Technical Report CS93-289, [|University of California, San Diego]

=Forum Posts=
 * [|Subject: Re: Game Learning] by Mike Gherrity, [|ai-repository], July 1, 1994
 * [|DB Tweaking Between Games] by Mike Gherrity, rgcc, May 13, 1997 » Kasparov versus Deep Blue 1997
 * [|Learning necessary for chess champion?] by Mike Gherrity, rgcc, May 16, 1997

=External Links= =References= =What links here?= include page="Michael Gherrity" component="backlinks" limit="40"
 * [|Michael Gherrity Home]
 * [|SAL] from [|Machine Learning in Games] by Jay Scott
 * [|The Mathematics Genealogy Project - Michael Gherrity]
 * Up one level**