Author: Dann Corbit
Date: 23:15:39 04/01/04
Go up one level in this thread
On April 02, 2004 at 01:55:46, Artem Pyatakov wrote: >Dann, > >A couple more simple questions in relation to your suggestions: > >>4. Book learning >This is learning that by playing a certain opening I (the engine) either won or >lost, correct? And using that to strike out certain lines in the future. The engine can do lots of things here. Typically, it will at least accumulate statistics for book positions. Sometimes, it will mark the last move in the book line if it turns out bad. Sometimes, it may even back up and analyze a line that turned out unexpected. The ce can also be stored. >>5. Position learning >What is this? Crafty does it. Maybe some others. You can also import the learning data from another machine output into crafty or even CAP data. >>6. TD-lambda learning >You just mean using this to tune the evaluation function, is that correct? >(KnightCap style) Yes. Several programs do this including KnightCap, BACE, and Prophet.
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.