From: "Brian Schröder" <ruby@...>
Date: 2004-12-11T05:15:05+09:00
Subject: Re: [QUIZ] Learning Tic-Tac-Toe (#11)

On Sat, 11 Dec 2004 04:54:04 +0900
James Edward Gray II <james@grayproductions.net> wrote:

> On Dec 10, 2004, at 1:42 PM, Brian Schr�der wrote:
> 
> > In this case, would a minimax-a/b algorithm be teaching the ai a 
> > strategie?
> > Tic-Tac-Toe is so shallow, it can be searched completely by 
> > minimax-a/b, so
> > this would be a one-page solution.
> 
> While you could argue that a minimax search has nothing to do with 
> Tic-Tac-Toe strategy, I doubt you would convince me on the evaluation 
> routine a minimax search requires.  Now if your program learns how to 
> evaluate positions over time...
> 
> James Edward Gray II
> 
> 

Well, the evalutation routine -100 for lost, +100 for won, 0 otherwise is not
really a evaluation routine, and minimax with alpha beta pruning can search the
whole problem in 6-7s on my machine.

But if I had to learn the rules, that would be more difficult...

Regards,

Brian

-- 
Brian Schr�der
http://ruby.brian-schroeder.de/