Author: Vasik Rajlich
Date: 02:19:30 04/01/04
Go up one level in this thread
On March 30, 2004 at 17:20:03, Jay Scott wrote: >On March 29, 2004 at 19:03:22, Uri Blass wrote: > >>On March 29, 2004 at 18:52:52, Jay Scott wrote: >> >>>Chess programmers, I've found, are remarkably resistant to changing how they do >>>certain traditional things. For example, every time I propose calibrating >>>evaluation not in millipawns but by some standard with a sounder theoretical >>>motivation, I'm shouted down by everyone who does not ignore me. I don't >>>understand that. >> >>I do not know what you mean but if you mean to evaluation in terms of expected >>result then >>I am not against evaluation not by pawns but by expected result. > >Yes, that's an example. > >>The problem is that it is not easy for human to think about defining the >>evaluation when it is in terms of expected result. > >Hmm, some humans, no doubt. Not the theoretically-minded, who are apparently a >small minority. But especially not when the evaluation is tuned by machine >learning, or even by a human using formal methods (imagine a test suite that >estimates how accurate each factor in the evaluator is). Then a sound >theoretical basis behind it all can help us come up with methods that are more >powerful. This is especially true when it comes to interactions between >evaluation features. > >Most chess programmers, apparently, find it easier or more satisfying to eyeball >their results and tune by the seat of the pants. > > Jay Hi, I would love to hear something a little more concrete. My own engine (Rybka) is too young to tune yet. I need to capture the right terms first. You don't want to just tune for example a passed pawn weight - first you need to discriminate the different types of passed pawns, then as a last step you determine the weights. However, I would be very curious to see which of my terms are doing anything. I do some tests disabling various terms, to see the magnitude of the differences in scores. However, this doesn't really say anything about how beneficial they are. Cheers, Vas
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.