[computer-go] MoGo paper at ICML
Hello all, We just presented our paper describing MoGo's improvements at ICML, and we thought we would pass on some of the feedback and corrections we have received. (http://www.machinelearning.org/proceedings/icml2007/papers/387.pdf) The way that we incorporate prior knowledge in UCT can be seen as a bayesian prior, and corresponds exactly to the dirichlet prior (more precisely to the beta prior as we here get binomials). The cumulative result is only given using the prior knowledge on top of RAVE, but it could have been done the other way round and give the same type of results. Each particular improvement is somehow independent of the others. On figure 5, the legend of horizontal axis should be m_prior rather than n_prior. All experiments (except the default policy) were played against GnuGo level 10, not level 8. Any other comments are welcome! Sylvain & David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Congratulations to GNU and to MoGoBot19!
On Sat, 2007-06-23 at 08:51 +0900, igo wrote: > I can understand what you are saying, > but still Fischer timing is not natural(simple) to my feeling. > Maybe because I am a fool. :-) > > There are many other time-limit competition's time system, > (football,boxing...) > I think they all have the same point: time goes steadily. (decreasing) > not like Fischer-timing, time goes irregularly. (increasing while > decreasing) > but I know this is a problem of feeling. Yes, I think so too. byo-yomi has something really odd, your clock can stand still even though you are using time. Unless you use too much, then suddenly you get penalized a whole chunk of time - otherwise none. This happens when the byo-yomi time is 60 seconds and as long as you use less than 60 seconds, it's as if you didn't use any time at all!How crazy is that? The difference between using 59.9 seconds and 60.1 seconds is about 300X greater than 60.1 - 59.9 - Don > Thanks a lot for the comments. > > igo > ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Congratulations to GNU and to MoGoBot19!
I can understand what you are saying, but still Fischer timing is not natural(simple) to my feeling. Maybe because I am a fool. :-) There are many other time-limit competition's time system, (football,boxing...) I think they all have the same point: time goes steadily. (decreasing) not like Fischer-timing, time goes irregularly. (increasing while decreasing) but I know this is a problem of feeling. Thanks a lot for the comments. igo - Don Dailey <[EMAIL PROTECTED]> wrote: > On Thu, 2007-06-21 at 19:38 +0900, igo wrote: > > Thank you very much. > > > > I think the writer was discussing from the "player's point of view", > > but not from GO's view. > > For the game of GO, if the time-system is fair and > > can avoid Sudden-Death naturally, that's enough. > > > > The writer's conclusion is "Keep it Simple!", > > but the Fischer timing's action is not simple to understand. > > If you are that locked into byo-yomi thinking you will probably think > anything else is "wrong." If you only know one thing it's simple and > everything else is complicated. > > I have an example of emotional sentimentality vs practical superiority > in the Post Script at the end of this. > > > Can someone explain me why a player receives times after played a move > > even when he doesn't lack of time ? > > Why does byo-yomi allocate a big chunk of time at the beginning even > though it's not needed?Would it be more "logical" from your point of > view to play the whole game in byo-yomi time so that you never have time > on the clock that is not needed? > > As I've stated before, you can think of any time-control system using > clocks in terms of who is given control over time-allocation. I think > it should be the players themselves, you lean in the direction of > thinking that decision should be automated and decided for the players. > > Sudden death is the least heavy-handed time-control system but fails to > recognize that it's impossible to predict how long a game might last. > So EVERY system is some compromise between you controlling your > time-allocation or having the control taken from you. > > A skilled human will do best when given as much control as possible over > his time allocation. However, it's probably also the case that an > inexperienced player will do better if those decisions are imposed on > him, like a child who needs the guidance of his parents before he is > mature enough to make wise decisions on his own. > > > If it's ok, instead of receiving times, I prefer receiving money. :-) > > Time is a far more precious resource than money. I'll take the time, > you can have the money :-) > > - Don > > > > P.S.I was playing club chess when the USCF decided to stop using > what is called "descriptive chess notation" and go with "algebraic > notation" which is what most of the world was then using and still > does. > > There was quite an outrage over this. There was no real pattern over > who embraced it and who didn't. Some very strong players resisted and > so did some very weak player and visa versa. As an observer of human > nature I tried to detect some pattern and what I thought I saw was two > different personality types. Some people were more into the "culture" > of chess (regardless of their strength) and others didn't care so much - > they just wanted to play chess. The ones who wanted to stay with the > archaic system also tended to know more about the history of chess and > in my (imperfect) judgment were the more intuitive type of players. Of > course there was nothing very scientific about this, it was based only > on my very subjective assessment.I was very fascinated with the > phenomenon and wondered why someone could be so passionate about > something like this.I immediately starting training myself to use > the new system and it slowed me down a little at first - it was awkward > getting used to something new when you are so comfortable with a > different way of doing things. > > It got harder and harder to get chess books that used descriptive > notation and over time virtually everyone stopped using it even for > their personal games. Just about everyone eventually acknowledged the > superiority of algebraic notation but it took a while for some. > > Personally, I expected the strong players to embrace the new system > (because I believed it was superior) and the weak players to be the > sentimental fools but it didn't seem to to work that way - there was no > pattern that I could detect in regard to the strength of players > advocating one system over the other. > > > - Don > > > > > igo > > > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http:/
Re: [computer-go] Congratulations to GNU and to MoGoBot19!
I can understand what you are saying, but still Fischer timing is not natural(simple) to my feeling. Maybe because I am a fool. :-) There are many other time-limit competition's time system, (football,boxing...) I think they all have the same point: time goes steadily. (decreasing) not like Fischer-timing, time goes irregularly. (increasing while decreasing) but I know this is a problem of feeling. Thanks a lot for the comments. igo - Don Dailey <[EMAIL PROTECTED]> wrote: > On Thu, 2007-06-21 at 19:38 +0900, igo wrote: > > Thank you very much. > > > > I think the writer was discussing from the "player's point of view", > > but not from GO's view. > > For the game of GO, if the time-system is fair and > > can avoid Sudden-Death naturally, that's enough. > > > > The writer's conclusion is "Keep it Simple!", > > but the Fischer timing's action is not simple to understand. > > If you are that locked into byo-yomi thinking you will probably think > anything else is "wrong." If you only know one thing it's simple and > everything else is complicated. > > I have an example of emotional sentimentality vs practical superiority > in the Post Script at the end of this. > > > Can someone explain me why a player receives times after played a move > > even when he doesn't lack of time ? > > Why does byo-yomi allocate a big chunk of time at the beginning even > though it's not needed?Would it be more "logical" from your point of > view to play the whole game in byo-yomi time so that you never have time > on the clock that is not needed? > > As I've stated before, you can think of any time-control system using > clocks in terms of who is given control over time-allocation. I think > it should be the players themselves, you lean in the direction of > thinking that decision should be automated and decided for the players. > > Sudden death is the least heavy-handed time-control system but fails to > recognize that it's impossible to predict how long a game might last. > So EVERY system is some compromise between you controlling your > time-allocation or having the control taken from you. > > A skilled human will do best when given as much control as possible over > his time allocation. However, it's probably also the case that an > inexperienced player will do better if those decisions are imposed on > him, like a child who needs the guidance of his parents before he is > mature enough to make wise decisions on his own. > > > If it's ok, instead of receiving times, I prefer receiving money. :-) > > Time is a far more precious resource than money. I'll take the time, > you can have the money :-) > > - Don > > > > P.S.I was playing club chess when the USCF decided to stop using > what is called "descriptive chess notation" and go with "algebraic > notation" which is what most of the world was then using and still > does. > > There was quite an outrage over this. There was no real pattern over > who embraced it and who didn't. Some very strong players resisted and > so did some very weak player and visa versa. As an observer of human > nature I tried to detect some pattern and what I thought I saw was two > different personality types. Some people were more into the "culture" > of chess (regardless of their strength) and others didn't care so much - > they just wanted to play chess. The ones who wanted to stay with the > archaic system also tended to know more about the history of chess and > in my (imperfect) judgment were the more intuitive type of players. Of > course there was nothing very scientific about this, it was based only > on my very subjective assessment.I was very fascinated with the > phenomenon and wondered why someone could be so passionate about > something like this.I immediately starting training myself to use > the new system and it slowed me down a little at first - it was awkward > getting used to something new when you are so comfortable with a > different way of doing things. > > It got harder and harder to get chess books that used descriptive > notation and over time virtually everyone stopped using it even for > their personal games. Just about everyone eventually acknowledged the > superiority of algebraic notation but it took a while for some. > > Personally, I expected the strong players to embrace the new system > (because I believed it was superior) and the weak players to be the > sentimental fools but it didn't seem to to work that way - there was no > pattern that I could detect in regard to the strength of players > advocating one system over the other. > > > - Don > > > > > igo > > > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http:/
