Re: [Computer-go] Seldom, but not alarming

[Ingo Althöfer]

> >Can you tell us the rules of the game? Maybe they help to explain the 
> >phenomenon. 
> 
> The game is https://boardgamegeek.com/boardgame/149910/six-making

Thanks. It might be an interesting test-stone for
MCTS procedures.
 
> The most unusual thing I see in the UCT tree is that at all the moves
> seem to be evaluated about the same, right up to the point where wins 
> start appearing; so the UCT tree is unusually uniform.  The typical
> branch factor is 40 or so, and all 40 moves tend to be have about the
> same number of visits and win rate...

> It's not a practical problem either, since the robot is devastatingly
> good compared to human players, even with very short time.

And even in longtime mode the bot is better?!
 
> If possible, make a PDF of your paper available.

Sorry, we transfered the copyright to Springer.
You will get a private mail from me ...

Ingo.
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Seldom, but not alarming

[Dave Dyer]

>Can you tell us the rules of the game? Maybe they help to explain the 
>phenomenon. 

The game is https://boardgamegeek.com/boardgame/149910/six-making

The most unusual thing I see in the UCT tree is that at all the moves
seem to be evaluated about the same, right up to the point where wins 
start appearing; so the UCT tree is unusually uniform.  The typical
branch factor is 40 or so, and all 40 moves tend to be have about the
same number of visits and win rate.

Currently, the best way to cope with the problem is to simply
stop the run when the node limit is hit. 

It's not a practical problem either, since the robot is devastatingly
good compared to human players, even with very short time.

If possible, make a PDF of your paper available.

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go