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

BGonline.org Forums

What do bots do to adjust their play ATS?

Posted By: Timothy Chow
Date: Saturday, 26 September 2009, at 5:51 p.m.

In Response To: What do bots do to adjust their play ATS? (Phil Simborg)

NJ, thanks for the info...that was very helpful.

Phil Simborg wrote:

Doesn't that mean that you cannot trust the gammon estimate for White?

Theoretically, the situation is even worse than that. Consider DMP. According to NJ, the bot will use a neural net that is trained on cubeless play to estimate the winning probabilities of various positions. But even if the neural net has a perfect understanding of cubeless play, this estimate will suffer from systematic error, because cubeless checker play differs from DMP checker play (gammons count in cubeless play). Thus DMP evaluations and rollouts will suffer from systematic error.

In practice these systematic errors may not cause many problems, but I imagine that the next generation of bots could be significantly improved by tuning neural nets specifically to play ATS.

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.