[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Why do bots have an even/odd effect?

Posted By: MaX
Date: Thursday, 20 August 2009, at 2:54 p.m.

In Response To: Why do bots have an even/odd effect? (Tom Keith)

Thanks Tom, so it's more or less what I was saying a few posts ago (http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=46264), with:

One thing I can figure out is the following: imagine a NN that systematically underesitmates the equity of the player on roll. A 0ply analysis with white on roll would then underestimate white's equity, but a 1ply analysis would overestimate white's equity. Of course this is not the case of gnubg. However, if it errs on a specific position and this position is N plies ahead and you analyse with exactly N plies, then you will "see" the error, while if you evaluate with more or less plies you may "avoid" the error.

So the so called "odd/even" effect is just the fact that in plied evals you can stumble on a postion that is (badly) misunderstood by the net. If the position "happens" only once, you will stumble on it only for a single ply setting, but if it repeats (like your example, persisting gap on pt.6) you may get right/wrong evals on odd/even plies.

That looks very much as a byproduct of the combination of "make 0ply as good as 2ply" and "we care only about relative equities" during learning.

I remeber somebody proposed (and coded) a "fractional ply" eval (e.g. 1.5 ply = average of 1ply and 2ply), but this may only mitigate the problem.

MaX.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.