Author: Robert Hyatt
Date: 10:36:02 03/22/03
Go up one level in this thread
On March 22, 2003 at 07:21:26, emerson tan wrote: >I heard that deep blue uses automatic tuning for its evaluation function, does >this mean that as it plays games against humans and computers, deep blue will >self tune its evaluation function base on the results of its games against >humans and computer? If it is, is it effective? Are the other programs using >automatic tuning also? We don't know much about this for Deep Blue. Deep Thought had an automated facility for tuning the eval, but it didn't do so when playing games. They had something like 10,000 selected positions and suggested best moves that they fed into the thing, and it used a least-squares minimization to tune the evaluation to prefer the best move after a shallow search. The Deep Blue and Deep Blue 2 evaluations were far more complex than deep thought and I never say anything specific about how they did the tuning there. I do remember Hsu talking about some sort of tool that could be used to manually tweak the eval for a specific case without having it break the evaluation for all the previous good cases they had worked on.
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.