Author: Gunnar Andersson
Date: 00:13:09 01/29/00
Go up one level in this thread
On January 28, 2000 at 20:50:13, Dann Corbit wrote: >On January 28, 2000 at 20:37:14, Tom Kerrigan wrote: >[snip] >>The only point that I'm trying to make is that the DB evaluation function isn't >>a mystical black box that does an infinite amount of work. (Some people think it >>is.) Its terms are probably very similar to a PC program's terms. Probably a PC >>program with extremely complicated/expensive terms, but still a PC program. > >And that point is a rather interesting point of discussion. However, since they >have several thousand tunable parameters, I suspect that it is a bit more >sophisticated than the average PC fare. > >How many such parameters are there in (for instance) Stobor [hardwired or >tunable]? > >The complexity of a system is a function of its inputs and outputs to some >degree. > >It is a miracle that they could debug something with that many inputs. It must >have been a nightmare. Depends a lot on what those inputs represented and how they were tuned. My Othello program Zebra has about 1'000'000 parameters in the evaluation function, tuned by solving a large optimization problem. Hsu et al might very well have done something similar. In fact, I have vague memories of a discussion on a hill-climbing algorithm they used already in the late 80s, for the first version of Deep Thought. The main problem then reduces to finding the right set of features. This is much simpler in e.g. Othello than in chess, but I'd be surprised if any chess program has thousands of features. / Gunnar
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.