Jonathan,
Hideki,
what strength player are you?
I'm IGS 4D
Just make your arguments, and others will realize your strength naturally.
in my experience:
strong players play good moves,
they dont play better moves in handicap games.
weak players play bad moves,
they dont play worse moves i
Hideki,
what strength player are you?
I'm IGS 4D
in my experience:
strong players play good moves,
they dont play better moves in handicap games.
weak players play bad moves,
they dont play worse moves in handicap games.
so... White effectively waits for Black to make two bad moves in a 2
s
Working through these ideas about wins, score, perfect play, etc... it is clear
that maximizing wins is the correct basic strategy. However, I still feel that
incorporating the score *somehow* should improve the winning estimates and
overall strength of MCTS.
See below for more thoughts.
>Fro
>2) Many go program authors have stated that "play to maximize wins"
>is stronger than "play to maximize points". I think this is because
>their evaluation functions are imperfectly optimistic--the program
>counts points that future play does not deliver.
You could be right, because really this i
Jonathan Chetwynd: :
>hideki wrote This is obviously wrong in handcap games,
>but what else is there?
>
>to start with the perhaps obvious,
>I believe komi is raised in pachi and tailed off as the game progresses.
>ie the goal is high to start and lowers as the game progresses.
>
>when playing as w
hideki wrote This is obviously wrong in handcap games,
but what else is there?
to start with the perhaps obvious,
I believe komi is raised in pachi and tailed off as the game progresses.
ie the goal is high to start and lowers as the game progresses.
when playing as white in the opening the goal
Interesting thoughts and I have a question.
How about handicap games? The opponent used in the simulations is self
in most (all?) MCTS programs. This is obviously wrong in handcap games
and the evaluation function returns wrong estimations of scores and
winning rates. So, the question is how
On Mon, Jul 4, 2011 at 11:02 AM, Ben Shoemaker wrote:
> >From: terry mcintyre
> >"The major one is that the MCTS scoring function is imperfect;
> historically, programs have snatched defeat from the jaws of victory by
> letting points be nibbled away in yose."
>
> (Apologies to those who understa
>From: terry mcintyre
>"The major one is that the MCTS scoring function is imperfect; historically,
>programs have snatched defeat from the jaws of victory by letting points be
>nibbled away in yose."
(Apologies to those who understand go and computer-go better than me--these are
just my thoug
2011/7/4 Andrés Domínguez :
> 2011/7/4 Erik van der Werf :
>>
>> If the program can play 30k level moves against you and still win its
>> a strong indication that you were playing a hopeless position. Sure
>> it's annoying, but that's because you should have resigned.
>
> I don't know if you are a
2011/7/4 Andrés Domínguez
> 2011/7/4 Erik van der Werf :
> >
> > If the program can play 30k level moves against you and still win its
> > a strong indication that you were playing a hopeless position. Sure
> > it's annoying, but that's because you should have resigned.
>
> I don't know if you ar
2011/7/4 Erik van der Werf :
>
> If the program can play 30k level moves against you and still win its
> a strong indication that you were playing a hopeless position. Sure
> it's annoying, but that's because you should have resigned.
I don't know if you are a Go player or not. If you are I suppos
2011/7/4 Andrés Domínguez :
> 2011/7/3 Álvaro Begué :
>> On Sun, Jul 3, 2011 at 8:36 AM, Leon Matoh wrote:
>>> [...]
>>> One of problems (which I tested with gogui, thankyou very much)
>>> was losing points in endgame when program is winning.
>>
>> Are you talking about positions where the program
2011/7/3 Andrés Domínguez
> 2011/7/3 Jean-loup Gailly :
> > Leon,
> >> One of problems (which I tested with gogui, thankyou very much)
> >> was losing points in endgame when program is winning.
> > This is by design. Pachi maximises the chance of winning, not the number
> > of points. But if you
e something for people to try!
>
> Terry McIntyre
>
> Unix/Linux Systems Administration
> Taking time to do it right saves having to do it twice.
>
>
> ------
> *From:* Álvaro Begué
>
> *To:* computer-go@dvandva.org
> *Sent:* Sun, Jul
This is what Many Faces of Go does. The traditional program adds a small
bias to the move selection, so when all else is equal it will play a more
reasonable looking move.
David
I think one solution that is used is to superimpose a more classical move
algorithm over this, so that by def
There are ways to make go programs play more like a human in this regard,
if you are willing to reduce the strength.You have to pick one,
maximum strength or some level of compromise to create the illusion of
playing stronger.There does not seem to be a way around this.
Here is a very s
>If the utility of any win is the same, it makes sense to
simply maximize the probability of winning. If we are not happy with
the program wasting points in a favorable endgame, it must be the case
that we are happier with a win by a large margin than with a win by a
small margin
I don't see a pro
_
From: Álvaro Begué
To: computer-go@dvandva.org
Sent: Sun, July 3, 2011 10:50:50 PM
Subject: Re: [Computer-go] MCTS and perfect endgame
On Sun, Jul 3, 2011 at 10:14 PM, terry mcintyre wrote:
> From: Jean-loup Gailly
> To: computer-go@dvandva.org
> Sent: Sun, July 3,
On Sun, Jul 3, 2011 at 10:14 PM, terry mcintyre wrote:
> From: Jean-loup Gailly
> To: computer-go@dvandva.org
> Sent: Sun, July 3, 2011 9:12:59 AM
> Subject: Re: [Computer-go] MCTS and perfect endgame
>
> Leon,
>> One of problems (which I tested with gogui, thankyou
>
> Why should this value be static? Shouldn't the behavior change when there
> is a certain win?
>
I think it should. This is what I do. When I have a win rate above 80% I
bias nodes by their average territory (counting prisoners). This might be
the same as adding a small bonus to each large win,
From: Jean-loup Gailly
To: computer-go@dvandva.org
Sent: Sun, July 3, 2011 9:12:59 AM
Subject: Re: [Computer-go] MCTS and perfect endgame
Leon,
> One of problems (which I tested with gogui, thankyou very much)
> was losing points in endgame when program is winning.
This is by design.
2011/7/3 Jean-loup Gailly :
> Leon,
>> One of problems (which I tested with gogui, thankyou very much)
>> was losing points in endgame when program is winning.
> This is by design. Pachi maximises the chance of winning, not the number
> of points. But if you want Pachi to win by more points while i
2011/7/3 Álvaro Begué :
> On Sun, Jul 3, 2011 at 8:36 AM, Leon Matoh wrote:
>> [...]
>> One of problems (which I tested with gogui, thankyou very much)
>> was losing points in endgame when program is winning.
>
> Are you talking about positions where the program losses a won game by
> not playing
Leon,
> One of problems (which I tested with gogui, thankyou very much)
> was losing points in endgame when program is winning.
This is by design. Pachi maximises the chance of winning, not the number
of points. But if you want Pachi to win by more points while increasing
the risk of losing, you
On Sun, Jul 3, 2011 at 8:36 AM, Leon Matoh wrote:
> [...]
> One of problems (which I tested with gogui, thankyou very much)
> was losing points in endgame when program is winning.
Are you talking about positions where the program losses a won game by
not playing the endgame correctly, or situatio
There is a Pachi mail list for users and developers:
http://rover.ms.mff.cuni.cz/mailman/listinfo/pachi
From: Leon Matoh
To: computer-go@dvandva.org
Sent: Sunday, July 3, 2011 7:36 AM
Subject: [Computer-go] MCTS and perfect endgame
MCTS and perfect endgame
Pachi is open source, so we can always look at the code you write. What's the
conceptual explanation of what you did?
Sent from my iPhone
On Jul 3, 2011, at 8:36 AM, Leon Matoh wrote:
> MCTS and perfect endgame
>
>
> Greetings.
>
> My name is Leon Matoh. I am go player, 5d, from Slovenia.
>
MCTS and perfect endgame
Greetings.
My name is Leon Matoh. I am go player, 5d, from Slovenia.
After long time I took time to work on go.
I downloaded couple of free programs and was surprised of
strength of pachi. Since it is open source i checked the code
and tested some hacks.
One of probl
29 matches
Mail list logo
