Author: Uri Blass
Date: 01:04:01 03/11/04
Go up one level in this thread
On March 11, 2004 at 03:08:36, José Carlos wrote: >On March 10, 2004 at 18:29:31, Uri Blass wrote: > >>On March 10, 2004 at 16:04:32, Dann Corbit wrote: >> >>>On March 10, 2004 at 14:41:47, Uri Blass wrote: >>> >>>>On March 10, 2004 at 14:23:29, Dave Gomboc wrote: >>>> >>>>>On March 09, 2004 at 16:05:15, Gian-Carlo Pascutto wrote: >>>>> >>>>>>Yet no top program does this, and they had a human correct it >>>>>>afterwards in Deep Blue. The conclusion should be obvious. >>>>> >>>>>Is that so? >>>>> >>>>>>If you can develop a *top level* evaluation function, better than >>>>>>good human tuning, solely on learning from a GM games database, >>>>>>you deserve an award. Nobody has succeeded before. >>>>> >>>>>Jonathan Schaeffer learned weights in Checkers [Chinook] without even using a >>>>>human games database (he used TD learning). The weights he tuned score 50% >>>>>against his hand-tuned code. >>>>> >>>>>I learned weights in Chess [Crafty] using 32k positions, hill-climbing an >>>>>ordinal correlation measure. It too scores 50% against the hand-tuned code. >>>> >>>>How many games and what time control? >>>>There is a difference if you score 50% with 2 games and with 2000 games? >>>> >>>>It is also possible that you get 50% against Crafty but less against other >>>>opponents. >>>> >>>> >>>>>Given Deep Sjeng's source code, I could zero its evaluation function weights, >>>>>and learn them from GM games to score 50% against the weights you have right now >>>>>too. >>>> >>>>You may be right but you cannot know about source code that you do not know. >>> >>>With his method, he will eventually reach a good result with any engine. >>>It uses generations, and discards the weaker ones absorbing the stronger ones. >>>After long enough waiting, it must become stronger. >> >>The question is how much time is long enough. >>It is clear that there is a way to find the best setting after enough time. >> >>Even the simple way of testing every possible setting can find the best setting >>if you only have 10^1000 years to wait. > > > The number of possible settings is infinite, so this method can't be sure to >find the best settings. > > José C. No The number of possible setting is finite but very big and if you have 200 parameters and everyone of them can get 10 values you have 10^200 setting. I admit that practically you probably have thousands of numbers when most of them can get more than 10 values so you have near 10^10000 possible setting so you nead probably at least 10^10000 years to find the best setting by testing all of them and 10^1000 is not enough. It does not change my point in this discussion that the question if you find the best setting if you wait enough time is not important and in this case waiting until programs solve chess may be shorter wait. Uri
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.