Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Mon, Feb 18, 2008 at 09:03:17PM +0100, Alain Baeckeroot wrote: Le lundi 18 février 2008, Michael Williams a écrit : But as was pointed out before, these high levels of MoGo are probably still not pro level, right? On 9x9 Big_slow_Mogo is near pro level, maybe more. 6 monthes ago or so it was able to regurlarly beat pro without komi on 9x9. But no komi on 9x9 is quite a handicap. It would be quite interesting to see how well the high level Mogos fare against high dans unhandicapped. -- Petr Pasky Baudis Whatever you can do, or dream you can, begin it. Boldness has genius, power, and magic in it.-- J. W. von Goethe ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Mon, 18 Feb 2008, Don Dailey wrote: Recently I have lost some faith in my belief that 7.0 komi is right on 9x9 with Chinese CGOS style rules. I was never absolutely SURE of it, but I believed it with a high degree of confidence. I still believe 7.0 is correct, but I'm somewhat less sure of this now. Doesn't the CGOS statistis you recently published show at slight advantage for WHITE at 7.5 komi? Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Don, Interesting thoughts and links. I read through them all. :) Some points: I wasn't expressing an opinion as to the degree of difference between God's komi and Man's komi. 2.5 seems perfectly reasonable (at least with current levels of skill). As far as it being widely believed that proper komi is independent of board size for all but the smallest boards -- I can't see a decent argument for assuming this. I can definitely see an argument that proper komi might *oscillate* around some target as boards get sufficiently huge, but it seems quite possible that the exact number of playable points on the board can result in some minor differences in score even as board sizes get very large, and it seems like it would take a rigorous proof for one to abandon that reasonable possibility. David On Feb 12, 2008, at 12:39 AM, Don Dailey wrote: David Schneider-Joseph wrote: On Feb 11, 2008, at 8:42 PM, Don Dailey wrote: David Schneider-Joseph wrote: On that topic - might it be possible that the notion of a proper komi, derived as it is from the hand of God (perfect play), will invariably be too high for any actual go players which would be an interesting match for each other? I guess it's possible. I don't think it's likely but I guess nobody can say with 100% certainty what the correct komi really is at any non-trivial board size. Why not likely? It seems a virtual guarantee to me. By definition, komi is proportional to the value of moving first. Likewise, by definition, your skill is the amount of value you get out of a move. Therefore, better players should play with higher komi. Hi David, It's possible (even easy) to construct positions where one side has a win, but the win requires careful accurate play or it loses. Such positions may actually be a practical advantage to the losing side if two equal players do not understand how to play it. The opening position in GO is such a position. I believe that if you pick the correct komi, whatever that may be, it's probably easier for white to win. This would imply an adjustment downward from god's number. This is essentially your argument and I agree with it. But how much adjustment?This is where we disagree. You seem to believe that the adjustment should be quite large. I disagree because even though I believe the white pieces are easier to play, I still believe that a won position is still an advantage for reasonably competent players. A strange consequence of your position is that you have to believe that a human player should prefer to start the game from a dead lost position. For instance if 12.0 is God's komi and 9.5 is man's komi, then 9.5 gives even chances in a position that is actually lost, and anything higher gives white a practical advantage in a dead lost position! Even though I believe as you do that it takes more skill to equalize with white (given the correct komi), I believe that 1/2 point more or less gives one side a winning game, and that is enough for players of modest skill to have the better winning chances. It's pretty clear however that white is easier to play ... If you play random vs random, 3.5 seems to be right komi.Since we both agree that komi should be AT LEAST 7.5, this implies that it's easier to play the white pieces for a player of limited skill (of course assuming komi is set correctly, whatever that may be.) And sure enough, if you use weak but not random program, the komi required jumps up very quickly. Even very weak programs seem to require about 7.5 komi, if they are beyond just weak beginner. But then even programs enormously stronger still require 7.5 komi. My feelings on this seem to match at least one source: Look here:http://senseis.xmp.net/?Komi Here is an excerpt: It is widely believed that the correct komi is independent of board size for all but the smallest boards. For area scoring, this would give 7 for 9x9+, 8 for 8x8, 7 for 7x7, 4 for 6x6, 25 for 5x5 (w cannot live), 0 for 4x4, 9 for 3x3, 4 for 2x2 with a superko rule, and 0 for 1x1. (these need to be verified) Despite all of this, I allowed the possibility that it's possible that even God cannot win at 7.5 komi. - Don ___ 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://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Le lundi 18 février 2008, Michael Williams a écrit : But as was pointed out before, these high levels of MoGo are probably still not pro level, right? On 9x9 Big_slow_Mogo is near pro level, maybe more. 6 monthes ago or so it was able to regurlarly beat pro without komi on 9x9. Alain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
? -Original Message- ? From: Don Dailey [EMAIL PROTECTED] ? To: computer-go computer-go@computer-go.org ? Sent: Mon, 18 Feb 2008 1:45 pm ? Subject: Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8 ...I have seen widely held beliefs be proven wrong before (the earth is flat is one example.) You are older than I thought. :-) On a more serious note, what is the status of the scalability study? - Dave Hillis More new features than ever. Check out the new AIM(R) Mail ! - http://webmail.aim.com ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
The next plan is to move to 13x13 with Mogo. We have failed to find a worthy second program so we will start with only mogo playing.Here is what we could use: 1. A strong scalable program. 2. Ability to adjust level in terms of number of play-outs. 3. binaries that work on 32 or 64 linux and macs. 4. Is not horribly resource intensive. The program doesn't have to be as strong as mogo, but it needs to be pretty strong as it is a huge waste of resources to test a program required 10X more cpu power to be equal in strength.I have asked some of the other really strong program authors in private and many people do not want to distribute a binary. - Don [EMAIL PROTECTED] wrote: -Original Message- From: Don Dailey [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Mon, 18 Feb 2008 1:45 pm Subject: Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8 ...I have seen widely held beliefs be proven wrong before (the earth is flat is one example.) You are older than I thought. :-) On a more serious note, what is the status of the scalability study? - Dave Hillis More new features than ever. Check out the new AIM(R) Mail http://o.aolcdn.com/cdn.webmail.aol.com/mailtour/aol/en-us/text.htm?ncid=aimcmp000501! ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Michael Williams wrote: But as was pointed out before, these high levels of MoGo are probably still not pro level, right? I don't know how strong Mogo is in the grand scheme of things - but the experiments with komi indicate that 7.0 is too low and that 8.0 is a lower bound on what komi should be (if we can take that experiment seriously.) The trend is that stronger players need higher komi. The other possibility we discussed is that perhaps it's easier to find the correct black moves early in the game than it is for white - in other words black has a practical advantage, but not a true advantage. - Don Don Dailey wrote: Hi David, Any opinion either of us have on this is only speculation. Nevertheless, in any kind of science there tends to be unproven conjectures that are widely believed to be true even though nobody has found a rigorous proof. Some of those will turn out to surprise everybody. I have seen widely held beliefs be proven wrong before (the earth is flat is one example.) Recently I have lost some faith in my belief that 7.0 komi is right on 9x9 with Chinese CGOS style rules. I was never absolutely SURE of it, but I believed it with a high degree of confidence. I still believe 7.0 is correct, but I'm somewhat less sure of this now. This is due to some testing I did with mogo, where at high levels mogo plays more even with 8.5 than 7.5. (black still wins slightly even at 8.5 komi) It's entirely possible, that mogo's style of play does better with black for whatever reason I don't understand.It could be some feature of the game, or even a mogo bug. I am reluctant to draw conclusions based on the performance of self play games of one computer player but this result has cast some doubt for me. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Maybe komi should be expressed in terms of percentage of the total number of positions. The komi of 7.5 for 9x9 looks the same with a komi of 7.5 for 19x19. But percentage wise, they are different. For a 1X1 board, the komi is 100% and for an infinitely large board, the komi is 0%. DL -Original Message- From: Don Dailey [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Mon, 18 Feb 2008 12:45 pm Subject: Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8 Hi David, Any opinion either of us have on this is only speculation. Nevertheless, in any kind of science there tends to be unproven conjectures that are widely believed to be true even though nobody has found a rigorous proof. Some of those will turn out to surprise everybody. I have seen widely held beliefs be proven wrong before (the earth is flat is one example.) Recently I have lost some faith in my belief that 7.0 komi is right on 9x9 with Chinese CGOS style rules. I was never absolutely SURE of it, but I believed it with a high degree of confidence. I still believe 7.0 is correct, but I'm somewhat less sure of this now. This is due to some testing I did with mogo, where at high levels mogo plays more even with 8.5 than 7.5. (black still wins slightly even at 8.5 komi) It's entirely possible, that mogo's style of play does better with black for whatever reason I don't understand.It could be some feature of the game, or even a mogo bug. I am reluctant to draw conclusions based on the performance of self play games of one computer player but this result has cast some doubt for me. - Don David Schneider-Joseph wrote: Don, Interesting thoughts and links. I read through them all. :) Some points: I wasn't expressing an opinion as to the degree of difference between God's komi and Man's komi. 2.5 seems perfectly reasonable (at least with current levels of skill). As far as it being widely believed that proper komi is independent of board size for all but the smallest boards -- I can't see a decent argument for assuming this. I can definitely see an argument that proper komi might *oscillate* around some target as boards get sufficiently huge, but it seems quite possible that the exact number of playable points on the board can result in some minor differences in score even as board sizes get very large, and it seems like it would take a rigorous proof for one to abandon that reasonable possibility. David On Feb 12, 2008, at 12:39 AM, Don Dailey wrote: David Schneider-Joseph wrote: On Feb 11, 2008, at 8:42 PM, Don Dailey wrote: David Schneider-Joseph wrote: On that topic - might it be possible that the notion of a proper komi, derived as it is from the hand of God (perfect play), will invariably be too high for any actual go players which would be an interesting match for each other? I guess it's possible. I don't think it's likely but I guess nobody can say with 100% certainty what the correct komi really is at any non-trivial board size. Why not likely? It seems a virtual guarantee to me. By definition, komi is proportional to the value of moving first. Likewise, by definition, your skill is the amount of value you get out of a move. Therefore, better players should play with higher komi. Hi David, It's possible (even easy) to construct positions where one side has a win, but the win requires careful accurate play or it loses. Such positions may actually be a practical advantage to the losing side if two equal players do not understand how to play it. The opening position in GO is such a position. I believe that if you pick the correct komi, whatever that may be, it's probably easier for white to win. This would imply an adjustment downward from god's number. This is essentially your argument and I agree with it. But how much adjustment?This is where we disagree. You seem to believe that the adjustment should be quite large. I disagree because even though I believe the white pieces are easier to play, I still believe that a won position is still an advantage for reasonably competent players. A strange consequence of your position is that you have to believe that a human player should prefer to start the game from a dead lost position. For instance if 12.0 is God's komi and 9.5 is man's komi, then 9.5 gives even chances in a position that is actually lost, and anything higher gives white a practical advantage in a dead lost position! Even though I believe as you do that it takes more skill to equalize with white (given the correct komi), I believe that 1/2 point more or less gives one side a winning game, and that is enough for players of modest skill to have the better winning chances. It's pretty clear however that white is easier to play ... If you play random vs random, 3.5 seems to be right komi.Since we both agree that komi should be AT LEAST 7.5, this implies
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Christoph Birk wrote: On Feb 11, 2008, at 9:39 PM, Don Dailey wrote: My feelings on this seem to match at least one source: Look here:http://senseis.xmp.net/?Komi Here is an excerpt: It is widely believed that the correct komi is independent of board size for all but the smallest boards. For area scoring, this would give 7 for 9x9+, 8 for 8x8, 7 for 7x7, 4 for 6x6, 25 for 5x5 (w cannot live), 0 for 4x4, 9 for 3x3, 4 for 2x2 with a superko rule, and 0 for 1x1. (these need to be verified) For 7x7 the komi is 9: http://senseis.xmp.net/?7x7BestPlay Yes, I agree. Someone should update that page. - Don Christoph ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
9.5pt komi is unreasonable. I agree with Don that perfect game value will probably turn out to be 7pts, though I'm keeping an open mind that it may be 6pts. I'd be surprised if it was 8pts, though that could just mean I've been analyzing the wrong openings :-). On 9x9 with Chinese rules even komi values (e.g. 6 or 8) are unlikely because they would require seki in all optimal lines of play. Consequently I'd expect 7, ... Interesting; I was thinking in terms of Japanese rules. It could turn out even the first move is different depending on the rule set! It'll be interesting to see when computers finally show it to us. Darren ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Tue, Feb 12, 2008 at 9:03 AM, Darren Cook [EMAIL PROTECTED] wrote: 9.5pt komi is unreasonable. I agree with Don that perfect game value will probably turn out to be 7pts, though I'm keeping an open mind that it may be 6pts. I'd be surprised if it was 8pts, though that could just mean I've been analyzing the wrong openings :-). On 9x9 with Chinese rules even komi values (e.g. 6 or 8) are unlikely because they would require seki in all optimal lines of play. Consequently I'd expect 7, but wouldn't completely rule out 5 and 9 just yet. Under Japanese rules I guess it can be anything from 5 to 10 (and maybe even the full range :-)). Erik ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Christoph Birk wrote: On Mon, 11 Feb 2008, Olivier Teytaud wrote: With 20 minute games, some people succeed in winning games against the release 3 of MoGo. But for X-hours-per-move, I don't know. What are the self-play results (white vs. black) for hour-long games of Mogo? I am wondering if the proper komi for 9x9 is really 7.5. Is your question whether 7.0 or 8.0 is the best komi? Or do you suspect a different 1/2 komi value is best? In thousands of tests I have run with strong programs, 7.5 appears to the best fractional komi. I believe 7.5 is a win for white with perfect play, but I'm not sure - however I am pretty sure it's the best fractional komi to use because 6.5 is clearly in blacks favor and 8.5 is clearly in whites favor. - Don Thanks, Christoph ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Mon, 11 Feb 2008, Olivier Teytaud wrote: With 20 minute games, some people succeed in winning games against the release 3 of MoGo. But for X-hours-per-move, I don't know. What are the self-play results (white vs. black) for hour-long games of Mogo? I am wondering if the proper komi for 9x9 is really 7.5. Thanks, Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Thinking a little more about it, I think we have to add an hypothesis which is that, for a given move, the number of AMAF updates if alpha (nb total UCT updates), with alpha 1. That seems to hold for most of the updates (with alpha close to 0.5), but there may be cases where it does not hold. If I understand well, you say that, in order to ensure consistency, we need some assumptions on the AMAF updates, i.e. the MC simulations which decide which move will have AMAF updates. Yes. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
I can't tell if you mean the float version or the double version. Using the float version (since it was all I had), I did a fairly extensive analysis of the losing move from the MoGo game that Fotland added comments to. My results were posted to this list on 2/1/08 under the subject, UCT and solving life and death. The test was run on 4 cores. Oops, I meant 2 threads (as stated in the original email). Thanks for your posts, the position is interesting. I'll try to see what happens with larger computation times. If it is a case of non-consistency it's interesting :-)I have no go-expertise to guess that this is the bad move from mogo, but I trust you for that :-) Well, if I find minutes for that after my fight with a furious myrinet switch :-) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Feb 11, 2008, at 9:39 PM, Don Dailey wrote: My feelings on this seem to match at least one source: Look here:http://senseis.xmp.net/?Komi Here is an excerpt: It is widely believed that the correct komi is independent of board size for all but the smallest boards. For area scoring, this would give 7 for 9x9+, 8 for 8x8, 7 for 7x7, 4 for 6x6, 25 for 5x5 (w cannot live), 0 for 4x4, 9 for 3x3, 4 for 2x2 with a superko rule, and 0 for 1x1. (these need to be verified) For 7x7 the komi is 9: http://senseis.xmp.net/?7x7BestPlay Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Andy wrote: But the program isn't stronger than pros, so how can it give better information about proper komi? Pro's cannot give you statistical information on komi unless you simply collate several thousand pro games. I don't think you need a particularly strong program, just good programs.If you notice that over thousands of games 6.5 is gives black a statistically significant edge, and 8.5 gives white a statistically significant edge, you know (at least for programs) that 8.5 is too high. Although it's possible that black has a won game at 8.5 komi, the evidence from computer play is just the opposite. You would have to assume that a computer is a better fighter when down, or conversely gets lazy when winning.Somehow that is difficult to believe. Also, you can try giving mogo a 6.5, and 8.5 komi and searching the second position (it seems to always play e5 on the first move.)At 6.5 komi, after black e5 white thinks it is slightly losing. At 8.5 komi white thinks it is slightly winning! At 7.5 komi it also thinks white is winning slightly. I tried Alford's value of 9.5 komi and white is even more happy, showing about 0.547 in the score. I don't believe what Alford says about 9.5 being the correct komi for 9x9.Where does that information come from? - Don On Feb 11, 2008 6:09 PM, Christoph Birk [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: On Mon, 11 Feb 2008, Don Dailey wrote: I don't bet, but if I did, I would bet that it's 7 or 8, and I'm fairly certain that with best play the game would end with 7 extra points for black. I think this was discussed at great length 2 or 3 years ago. I know ... I brought it up again because of Mogo's success. A very (!) strong program should be able to tell us the proper komi. Christoph ___ computer-go mailing list computer-go@computer-go.org mailto:computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
But the program isn't stronger than pros, so how can it give better information about proper komi? On Feb 11, 2008 6:09 PM, Christoph Birk [EMAIL PROTECTED] wrote: On Mon, 11 Feb 2008, Don Dailey wrote: I don't bet, but if I did, I would bet that it's 7 or 8, and I'm fairly certain that with best play the game would end with 7 extra points for black. I think this was discussed at great length 2 or 3 years ago. I know ... I brought it up again because of Mogo's success. A very (!) strong program should be able to tell us the proper komi. Christoph ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Christoph Birk wrote: On Mon, 11 Feb 2008, Don Dailey wrote: Is your question whether 7.0 or 8.0 is the best komi? Or do you suspect a different 1/2 komi value is best? I wonder what the true komi is ... I don't know (nobody knows?) if it's fractional or not; eg. for 7x7 it is 9.0. I think the true komi must be an even number. For CGOS, the question is what is the best komi since drawn games are simply not allowed. I don't bet, but if I did, I would bet that it's 7 or 8, and I'm fairly certain that with best play the game would end with 7 extra points for black. I think this was discussed at great length 2 or 3 years ago. - Don Christoph ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Mon, 11 Feb 2008, Don Dailey wrote: Is your question whether 7.0 or 8.0 is the best komi? Or do you suspect a different 1/2 komi value is best? I wonder what the true komi is ... I don't know (nobody knows?) if it's fractional or not; eg. for 7x7 it is 9.0. Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Olivier Teytaud: [EMAIL PROTECTED]: That translates to mean that MoGo no longer uses upper confidence bounds, and only uses means. It also means that MoGo will _never_ explore improbable children (after a few sims) unless the RAVE value yields an unusually high estimate for it. Is all of that correct? Precisely: I don't see why you would be wrong, but empirically for 9x9, we have played games against high-level humans and for the (few :-) ) games that mogo lost, we tried to see which moves were erroneously chosen by mogo; if we restart mogo at the same position with a huge computation time (30 minutes of a fast octocore) mogo always changed his mind and moves to a better move. Could we look at some of the records of the games? -Hideki So: - theoretically, I don't see any reason for mogo to be asymptotically consistent - there are long computation times during which mogo focuses on a bad move - however, we have not seen a case of bad move for which mogo keeps this move in case of _very_ long computation times == if someone beats the release MoGoR3 with very large computation times (time x nbcores = 4h, 1 to 4 cores) I'm interested in the sgf file and the analysis ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Quoting Don Dailey [EMAIL PROTECTED]: My feelings on this seem to match at least one source: Look here:http://senseis.xmp.net/?Komi Here is an excerpt: It is widely believed that the correct komi is independent of board size for all but the smallest boards. For area scoring, this would give 7 for 9x9+, 8 for 8x8, 7 for 7x7, 4 for 6x6, 25 for 5x5 (w cannot live), 0 for 4x4, 9 for 3x3, 4 for 2x2 with a superko rule, and 0 for 1x1. (these need to be verified) I corrected this sensei page to give komi 9 for 7x7 and added a link to the sgf file John Tromp provides with the analysis. I played a lot with Valkyria on 7x7 and although it proofs nothing it is really convincing that 9 is the correct komi. Similarily one find that in very simple games on 9x9, but where the moves are good solid shape white almost always win with 0.5 points with 7.5 komi. Thus if one designs and opening book for 9x9 (as I tried to) one should try to complicate things as black and play simple as white early on. The exception may be when black opens at 5,5 in the center. Then often white ends up trying to live with two groups which can be very difficult against a competent opponent. -Magnus ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Feb 12, 2008 2:10 PM, Don Dailey [EMAIL PROTECTED] wrote: Andy wrote: But the program isn't stronger than pros, so how can it give better information about proper komi? Pro's cannot give you statistical information on komi unless you simply collate several thousand pro games. I don't think you need a particularly strong program, just good programs.If you notice that over thousands of games 6.5 is gives black a statistically significant edge, and 8.5 gives white a statistically significant edge, you know (at least for programs) that 8.5 is too high. I think you might need a strong program with either (a) no built-in knowledge about the game of go (i.e. pure UCT with no open book, no heuristics, etc) or (b) with built-in knowledge which can be shown to be of equal benefit to both black and white. I'm guessing that (a) will happen before (b). cheers stuart ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Feb 11, 2008, at 8:42 PM, Don Dailey wrote: David Schneider-Joseph wrote: On that topic - might it be possible that the notion of a proper komi, derived as it is from the hand of God (perfect play), will invariably be too high for any actual go players which would be an interesting match for each other? I guess it's possible. I don't think it's likely but I guess nobody can say with 100% certainty what the correct komi really is at any non-trivial board size. Why not likely? It seems a virtual guarantee to me. By definition, komi is proportional to the value of moving first. Likewise, by definition, your skill is the amount of value you get out of a move. Therefore, better players should play with higher komi. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
David Schneider-Joseph wrote: On that topic - might it be possible that the notion of a proper komi, derived as it is from the hand of God (perfect play), will invariably be too high for any actual go players which would be an interesting match for each other? I guess it's possible. I don't think it's likely but I guess nobody can say with 100% certainty what the correct komi really is at any non-trivial board size. - Don On Feb 11, 2008, at 7:35 PM, Andy wrote: But the program isn't stronger than pros, so how can it give better information about proper komi? On Feb 11, 2008 6:09 PM, Christoph Birk [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: On Mon, 11 Feb 2008, Don Dailey wrote: I don't bet, but if I did, I would bet that it's 7 or 8, and I'm fairly certain that with best play the game would end with 7 extra points for black. I think this was discussed at great length 2 or 3 years ago. I know ... I brought it up again because of Mogo's success. A very (!) strong program should be able to tell us the proper komi. Christoph ___ computer-go mailing list computer-go@computer-go.org mailto:computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org mailto:computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On that topic - might it be possible that the notion of a proper komi, derived as it is from the hand of God (perfect play), will invariably be too high for any actual go players which would be an interesting match for each other? On Feb 11, 2008, at 7:35 PM, Andy wrote: But the program isn't stronger than pros, so how can it give better information about proper komi? On Feb 11, 2008 6:09 PM, Christoph Birk [EMAIL PROTECTED] wrote: On Mon, 11 Feb 2008, Don Dailey wrote: I don't bet, but if I did, I would bet that it's 7 or 8, and I'm fairly certain that with best play the game would end with 7 extra points for black. I think this was discussed at great length 2 or 3 years ago. I know ... I brought it up again because of Mogo's success. A very (!) strong program should be able to tell us the proper komi. Christoph ___ 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://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Don Dailey wrote: I tried Alford's value of 9.5 komi and white is even more happy, showing about 0.547 in the score. I don't believe what Alford says about 9.5 being the correct komi for 9x9.Where does that information come from? Japanese tv games. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Mon, 11 Feb 2008, Don Dailey wrote: I don't bet, but if I did, I would bet that it's 7 or 8, and I'm fairly certain that with best play the game would end with 7 extra points for black. I think this was discussed at great length 2 or 3 years ago. I know ... I brought it up again because of Mogo's success. A very (!) strong program should be able to tell us the proper komi. Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Christoph Birk wrote: On Mon, 11 Feb 2008, Don Dailey wrote: Is your question whether 7.0 or 8.0 is the best komi? Or do you suspect a different 1/2 komi value is best? I wonder what the true komi is ... I don't know (nobody knows?) if it's fractional or not; eg. for 7x7 it is 9.0. Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ unlurk i believe correct komi for 9x9 with pros is 9.5 lurk ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
As far as I see, if RAVE gives constant value 0 to one move, it will never be tested if other moves have non-zero AMAF values. A move with real empirical probability 0 of winning and AMAF value of 0.01 will always be preferred to a non-simulated move with AMAF 0.0, whatever may be the number of simulations. I agree, it is why I added a statement about the prior, which implies that the AMAF value is never 0.0 but at worst decreases like 1/m if m is the number of AMAF updates for that move. Thinking a little more about it, I think we have to add an hypothesis which is that, for a given move, the number of AMAF updates if alpha (nb total UCT updates), with alpha 1. That seems to hold for most of the updates (with alpha close to 0.5), but there may be cases where it does not hold. Maybe I am confused and say unsound things, sorry for that. It is the kind of things we should discuss in front of a black (or white) board. Sylvain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Sylvain wrote: Thinking a little more about it, I think we have to add an hypothesis which is that, for a given move, the number of AMAF updates if alpha (nb total UCT updates), with alpha 1. That seems to hold for most of the updates (with alpha close to 0.5), but there may be cases where it does not hold. If I understand well, you say that, in order to ensure consistency, we need some assumptions on the AMAF updates, i.e. the MC simulations which decide which move will have AMAF updates. (this would be a good piece of news for the two people trying to get rid of some bias in the MC :-) unfortunately, it is difficult to have statistics on the level of mogo depending on that. It it gives some result, it is probably only for huge computation times and very specific positions... we need 100 000 years before ensuring that with 5% confidence intervals on complete games :-) ) (in empirical cases, I'll try to check the consistency on the example posted by David Michael) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
A new position is always visited unless the leaf of the tree is the end of the game. In that case, one player always win, so the other always win. Then, the losing player will explore all the other moves to avoid the sure loss. If all moves are still loosing, that will propagate to the move before, and the exploration will begin and so on. (min -- loss I guess) As far as I see, if RAVE gives constant value 0 to one move, it will never be tested if other moves have non-zero AMAF values. A move with real empirical probability 0 of winning and AMAF value of 0.01 will always be preferred to a non-simulated move with AMAF 0.0, whatever may be the number of simulations. So, I don't see why the bandit would be consistent, unless we have assumptions on the MC or on RAVE values. I might be completly wrong, as I said I have only retro-engineered the bandit in mogo until the recent PDF file. I trust your opinion more than mine :-) There are people studying some specific positions with surprising behavior, but I am not working on that with them, they might want to post their analysis in this mailing-list... ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Mon, 11 Feb 2008, Michael Alford wrote: i believe correct komi for 9x9 with pros is 9.5 That's way too large. Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
David Schneider-Joseph wrote: On Feb 11, 2008, at 8:42 PM, Don Dailey wrote: David Schneider-Joseph wrote: On that topic - might it be possible that the notion of a proper komi, derived as it is from the hand of God (perfect play), will invariably be too high for any actual go players which would be an interesting match for each other? I guess it's possible. I don't think it's likely but I guess nobody can say with 100% certainty what the correct komi really is at any non-trivial board size. Why not likely? It seems a virtual guarantee to me. By definition, komi is proportional to the value of moving first. Likewise, by definition, your skill is the amount of value you get out of a move. Therefore, better players should play with higher komi. Hi David, It's possible (even easy) to construct positions where one side has a win, but the win requires careful accurate play or it loses. Such positions may actually be a practical advantage to the losing side if two equal players do not understand how to play it. The opening position in GO is such a position. I believe that if you pick the correct komi, whatever that may be, it's probably easier for white to win. This would imply an adjustment downward from god's number. This is essentially your argument and I agree with it. But how much adjustment?This is where we disagree. You seem to believe that the adjustment should be quite large. I disagree because even though I believe the white pieces are easier to play, I still believe that a won position is still an advantage for reasonably competent players. A strange consequence of your position is that you have to believe that a human player should prefer to start the game from a dead lost position. For instance if 12.0 is God's komi and 9.5 is man's komi, then 9.5 gives even chances in a position that is actually lost, and anything higher gives white a practical advantage in a dead lost position! Even though I believe as you do that it takes more skill to equalize with white (given the correct komi), I believe that 1/2 point more or less gives one side a winning game, and that is enough for players of modest skill to have the better winning chances. It's pretty clear however that white is easier to play ... If you play random vs random, 3.5 seems to be right komi.Since we both agree that komi should be AT LEAST 7.5, this implies that it's easier to play the white pieces for a player of limited skill (of course assuming komi is set correctly, whatever that may be.) And sure enough, if you use weak but not random program, the komi required jumps up very quickly. Even very weak programs seem to require about 7.5 komi, if they are beyond just weak beginner. But then even programs enormously stronger still require 7.5 komi. My feelings on this seem to match at least one source: Look here:http://senseis.xmp.net/?Komi Here is an excerpt: It is widely believed that the correct komi is independent of board size for all but the smallest boards. For area scoring, this would give 7 for 9x9+, 8 for 8x8, 7 for 7x7, 4 for 6x6, 25 for 5x5 (w cannot live), 0 for 4x4, 9 for 3x3, 4 for 2x2 with a superko rule, and 0 for 1x1. (these need to be verified) Despite all of this, I allowed the possibility that it's possible that even God cannot win at 7.5 komi. - Don ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Sat, 2008-02-09 at 11:50 +0100, Olivier Teytaud wrote: I think it is time to share this idea with the world :-) The idea is to estimate bias and variance to calculate the best combination of UCT and RAVE values. I have attached a pdf explaining the new formula. It is written in the pdf file that the formula is the one in MoGo; but in MoGo there's no sqrt(log...), i.e. c_r=c_u=0. That translates to mean that MoGo no longer uses upper confidence bounds, and only uses means. It also means that MoGo will _never_ explore improbable children (after a few sims) unless the RAVE value yields an unusually high estimate for it. Is all of that correct? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Sun, 2008-02-10 at 18:35 +0100, Olivier Teytaud wrote: That translates to mean that MoGo no longer uses upper confidence bounds, and only uses means. It also means that MoGo will _never_ explore improbable children (after a few sims) unless the RAVE value yields an unusually high estimate for it. Is all of that correct? Precisely: I don't see why you would be wrong, but empirically for 9x9, we have played games against high-level humans and for the (few :-) ) games that mogo lost, we tried to see which moves were erroneously chosen by mogo; if we restart mogo at the same position with a huge computation time (30 minutes of a fast octocore) mogo always changed his mind and moves to a better move. I'm just surprised to hear that the program that introduced UCT (and got so many others to use it), isn't using UCT any more. Combining RAVE and UCT as described in the PDF still sounds like UCT to me, but with no sqrt(log) term, it no longer is. I'll certainly have to think about the trades being made and what I'd expect the outcome to be. 4 hours of CPU time seems like a really long time. Have there been any trades done to measure how long MoGo takes to change its mind under different configurations? So: - theoretically, I don't see any reason for mogo to be asymptotically consistent Asymptotically approaching perfect play is no longer a goal? (Rhetorical question to show some concern) - there are long computation times during which mogo focuses on a bad move - however, we have not seen a case of bad move for which mogo keeps this move in case of _very_ long computation times == if someone beats the release MoGoR3 with very large computation times (time x nbcores = 4h, 1 to 4 cores) I'm interested in the sgf file and the analysis I bet someone will take up the challenge. You should probably also give a limit to how long the human can think about the game. Can they also sit there for 4 hours contemplating a move and working out all the variations? Of course, this challenge by itself shows the ultimate effect of increased computational power... ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
I'm just surprised to hear that the program that introduced UCT (and got so many others to use it), isn't using UCT any more. Combining RAVE and UCT as described in the PDF still sounds like UCT to me, but with no sqrt(log) term, it no longer is. I'll certainly have to think about the trades being made and what I'd expect the outcome to be. I agree that this is Bandit Based Monte-Carlo planning much more than UCT. When the RAVE paper has been published, the UCT term was present, but I think Sylvain can confirm that it has been removed a short time later - this has been posted in the mailing list a long time ago. (no extensive study yet of the precise computation time required by mogo for removing all bad moves - we are mainly analyzing _if_ some moves are still weak in spite of large computation times - this is already quite hard, as guessing which move(s) is (are) bad is not easy - finally, this is a 9x9 study with humans in the loop!) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
This really interesting. The strongest Mogo no longer uses UCT at all. It's tuned instead to do very deep and narrow searches. I've seen that other programs that use UCT are using very small C values to make uct also do very narrow searches. The strong programs also have very smart playouts. With the patterns they use, I expect that when they make a random choice it is between a very small number of options. Can we even call these programs UCT/MC any more? They are not using the bandit problem UCT search any more. They are more like greedy best-first searchers that keep going deeper in the variation with highest winning probability. David -Original Message- From: [EMAIL PROTECTED] [mailto:computer-go- [EMAIL PROTECTED] On Behalf Of Olivier Teytaud Sent: Sunday, February 10, 2008 9:35 AM To: computer-go Subject: Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8 That translates to mean that MoGo no longer uses upper confidence bounds, and only uses means. It also means that MoGo will _never_ explore improbable children (after a few sims) unless the RAVE value yields an unusually high estimate for it. Is all of that correct? Precisely: I don't see why you would be wrong, but empirically for 9x9, we have played games against high-level humans and for the (few :-) ) games that mogo lost, we tried to see which moves were erroneously chosen by mogo; if we restart mogo at the same position with a huge computation time (30 minutes of a fast octocore) mogo always changed his mind and moves to a better move. So: - theoretically, I don't see any reason for mogo to be asymptotically consistent - there are long computation times during which mogo focuses on a bad move - however, we have not seen a case of bad move for which mogo keeps this move in case of _very_ long computation times == if someone beats the release MoGoR3 with very large computation times (time x nbcores = 4h, 1 to 4 cores) I'm interested in the sgf file and the analysis ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
- at each (or every n) iteration you add one node. As far as I see, new nodes are created only if new nodes are visited; if score(visited nodes) score(unvisited nodes) why would mogo visit new nodes ? But (before the recent PDF file) I never understood completly the bandit in mogo, so you are probably right :-) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
A new position is always visited unless the leaf of the tree is the end of the game. In that case, one player always win, so the other always win. Then, the losing player will explore all the other moves to avoid the sure loss. If all moves are still loosing, that will propagate to the move before, and the exploration will begin and so on. There is indeed no forced exploration, but there is exploration as soon as a move is loosing. I can totally be wrong, but currently I don't see where this does not hold. Does it? Sylvain 2008/2/10, Olivier Teytaud [EMAIL PROTECTED]: - at each (or every n) iteration you add one node. As far as I see, new nodes are created only if new nodes are visited; if score(visited nodes) score(unvisited nodes) why would mogo visit new nodes ? But (before the recent PDF file) I never understood completly the bandit in mogo, so you are probably right :-) ___ 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://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
== if someone beats the release MoGoR3 with very large computation times (time x nbcores = 4h, 1 to 4 cores) I'm interested in the sgf file and the analysis I can't tell if you mean the float version or the double version. Using the float version (since it was all I had), I did a fairly extensive analysis of the losing move from the MoGo game that Fotland added comments to. My results were posted to this list on 2/1/08 under the subject, UCT and solving life and death. The test was run on 4 cores. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
I can't tell if you mean the float version or the double version. Using the float version (since it was all I had), I did a fairly extensive analysis of the losing move from the MoGo game that Fotland added comments to. My results were posted to this list on 2/1/08 under the subject, UCT and solving life and death. The test was run on 4 cores. Float or double does not matter a lot, but the computation time matters. With 20 minute games, some people succeed in winning games against the release 3 of MoGo. But for X-hours-per-move, I don't know. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
Michael Williams wrote: == if someone beats the release MoGoR3 with very large computation times (time x nbcores = 4h, 1 to 4 cores) I'm interested in the sgf file and the analysis I can't tell if you mean the float version or the double version. Using the float version (since it was all I had), I did a fairly extensive analysis of the losing move from the MoGo game that Fotland added comments to. My results were posted to this list on 2/1/08 under the subject, UCT and solving life and death. The test was run on 4 cores. Oops, I meant 2 threads (as stated in the original email). ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
David Silver wrote: I think it is time to share this idea with the world :-) Great. Thanks for sharing. Rémi ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: computer-go Digest, Vol 43, Issue 8
Thank you very much, Silver. Interesting report! -Hideki David Silver: [EMAIL PROTECTED]: Hi all, On 7-Feb-08, at 1:30 AM, [EMAIL PROTECTED] wrote: Note as well that the current implementation of MoGo (not the one at the time of the ICML paper) use a different tradeoff between UCT and Rave value, thanks to an idea of David Silver, which brought improvements in 19x19 (where the Rave values are the most useful), while it was marginal (still better) in 9x9. But anyway we here are talking about 9x9, so it can't explain what you are talking about. I think it is time to share this idea with the world :-) The idea is to estimate bias and variance to calculate the best combination of UCT and RAVE values. I have attached a pdf explaining the new formula. inline file (2) () Depending on the playout policy, adding an upper confidence bound to the rave values can push some terrible bad moves up (like playing on 1-1). The reason seems to be that such moves are normally sampled very infrequently (so the UCB will be higher), and when they are selected (...) That could be an explanation, but there are two points: - the prior you put on top of Rave often avoid to first sample 1-1, and even when you do, you very often loose just 1 playout because of the UCT value you get right away. - I never observed a big discrepancy between the number of Rave samples for each move. Also, the upper confidence bound reduces rapidly with RAVE, because so many moves are played in each playout. So even without prior knowledge, moves like the 1-1 point should be observed less when using RAVE, because they will quickly become associated with losing games. RAVE acts like a pruning mechanism - these bad moves don't even need to be played in the tree, to identify that they are a bad idea. It is also like progressive widening, because all moves are tried in the tree eventually, once the UCT estimate starts to dominate the RAVE estimate. So it is perhaps not a surprise that programs with pruning and progressive widening see less improvement when implementing RAVE - the ideas overlap a great deal. Of course, the all-moves-as-first heuristic is often wrong - so RAVE can make big mistakes. But on average it improves performance, which is what matters. -Dave inline file ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: computer-go Digest, Vol 43, Issue 8
On Feb 8, 2008 12:09 PM, David Silver [EMAIL PROTECTED] wrote: I think it is time to share this idea with the world :-) The idea is to estimate bias and variance to calculate the best combination of UCT and RAVE values. I have attached a pdf explaining the new formula. Thanks! The original paper's formula for beta always felt wrong to me. I like this new one a lot better. Is it correct that the pdf assumes a uct bias of zero? Calculation of the MSE seems to assume this going into the last step but doesn't simplify life by doing it in the first reduction... Maybe it's just academic, but when I plug in bias = 0, I don't get the UCT formula for sims = n+m. Q comes out correct, but Q+ does not. I guess I'd sort of expect to see something along the lines of Q+ur = Qur + c*sqrt(log(???)/x) where x = B^2/m + (1-B)^2/n. When br = 0, x reduces to m+n. Maybe I'm just crazy and there's no good way to compute ??? inside my log. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
