Author: Stuart Cracraft
Date: 06:45:42 11/01/04
Go up one level in this thread
On November 01, 2004 at 04:48:24, GuyHaworth wrote: > >Why not ask GT directr? > >http://www.research.ibm.com/massive/tdl.html#about_the_author Well -- the reason is that I have found it very rare that an author will give out something free that he is essentially charging for. > >My understanding is that ANNs work better for backgammon than for games which do >not have the 'smoothing' effect (on the position evaluation function) caused by >randomness. > There are ways around this. > >Meel Velliste and I ran an ANN-experiment on KPK some years back, somewhat >helped by GT's papers. > >g More than that. GT is absolutely one of the most creative forces to hit strategic games. He has showed, beyond TD-Gammon what he did for Deep Blue. The move that Gary Kasparov called "plays like God" in regards to Deep Blue I believe would not have been made about Tesauro's evaluation tuning effort for DB.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.