[ 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: Tom Keith
Date: Monday, 17 August 2009, at 11:16 p.m.

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

Hi MaX,

Here is a made-up example. Suppose a NN has a defect where it wrongly overvalues any position in which the side to roll has a gap on their own 6-point. Look at what happens when we run into such a position.

Suppose A has a gap on his 6-point and let's say the true value of A's position is 0.63. At 0-ply, the NN will incorrectly report A's equity as 0.73 because of the defect described above.

At 1-ply, the lookahead routine takes a roll for A and plays it. B is now on roll, and the position is evaluated from B's point of view. B doesn't have a 6-point gap so the NN correctly evaluates the position from B's point of view. It does the same (correct) evaluation for each of the 21 possible rolls, computes a weighted average, inverts the equity, and returns the correct result of 0.63.

So 0-ply reports 0.73 but 1-ply reports 0.63. This alternation between even and odd plies will persist as long as A continues to have a gap on his 6-point. It happens because, on even-ply evaluations, the position passed to the NN is always from A's point of view. And on odd ply evaluations, the position passed to the NN is always from B's point of view.

Does that help?

Tom

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.