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

BGonline.org Forums

this should be easy.. - ROLLOUT

Posted By: Tom Keith
Date: Saturday, 28 June 2014, at 2:12 p.m.

In Response To: this should be easy.. - ROLLOUT (Timothy Chow)

I'll take your word for it that those values (which I believe no bot has ever computed, or even tried to compute, at a match score where the cube is live) are very useful, even though I don't see why.

I'm sure bot makers have tried to compute these numbers. Traditionally, at least, bot neural nets are trained using cubeless money play. When you feed a position into a NN you get back six numbers: the probability of winning a single game, gammon, and backgammon, and losing a single game gammon, and backgammon. In a match situation the bot has to compute a "value vector" to apply to the cubeless values. Take the inner product of the value vector with the cubeless numbers produced by the NN and you have an estimate of the match equity.

Modern bots may even do better than this. Once you have an accurate value vector, you can feed it as input to the bot's NN. Now the bot is truly playing according to the score.

Knowing these numbers is valuable for human players too. How hard should you try to win (or avoid) a gammon or backgammon?

I have a model, which I believe is quite accurate, that calculates a value vector for any match situation. (I have always planned to write an article about it but haven't got around to it (yet).)

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.