| Life In 19x19 http://www.lifein19x19.com/ |
|
| The just over Olympiad http://www.lifein19x19.com/viewtopic.php?f=18&t=2099 |
Page 1 of 1 |
| Author: | Mike Novack [ Tue Oct 12, 2010 9:30 am ] |
| Post subject: | The just over Olympiad |
Results? Does anybody know where to look to see how this turned out? Early on the AGA main page said MFOG won the 13x13 and that there would be more news forthcoming but since then nada. |
|
| Author: | oren [ Tue Oct 12, 2010 10:26 am ] |
| Post subject: | Re: The just over Olympiad |
http://www.grappa.univ-lille3.fr/icga/event.php?id=42 Somewhat off topic, Kanazawa is a great city. |
|
| Author: | Mike Novack [ Tue Oct 12, 2010 2:17 pm ] |
| Post subject: | Re: The just over Olympiad |
Thanks |
|
| Author: | Stable [ Tue Oct 12, 2010 6:21 pm ] |
| Post subject: | Re: The just over Olympiad |
I've just been playing the 9x9 winner mygofriend on KGS as an anti-insomia-boredom-thing. It's wiped the floor with me 2ce. I nearly had it the last time, but then it made a rather clever overplay and tricked me into losing by 0.5! I'm intruiged because they're based in London. Anyone know Frank Karger? |
|
| Author: | Mike Novack [ Wed Oct 13, 2010 3:26 pm ] |
| Post subject: | Re: The just over Olympiad |
Folks need to keep in mind that the performance of these programs is a combination of the program itself, how much computing power available, and the time control. As you can see if you follow that link Oren provided, at the Olympiad many of these bots were running on much more "iron" than you or I have available at home. And it's not linear. For each of these programs there is an amount of time + crunch power at which it will perform "adequately" but below which rather poorly. Providing more crunch power and/or time will let them play stronger but not necessarily all that much stronger (diminishing returns means above some point a very shallow improvement curve). And this point is going to be different for each of the programs. Always try to find out "how powerful a machine is that bot running on that I am playing against on the server?" and especially find that out if it's commercial software you are considering buying (you need to know how strong it wil be on your computer). In each case (for each program). For example, here we had Erica and MFOG running on comparable machines with Zen a couple times more machine and coming out roughly tied (each lost to one of the others and beat one of the others and won all their other games). Some of the other machines in this event were much more powerful. However, you should not assume that the same thing would have happened were all programs limited to say 2 core 2 gHz machines (a good modern laptop, but not a high end "mobile workstation"). Yes you could expect Erica, Zen, and MFOG would beat the other 5 but no, they might no longer be roughly tied when limited to less machine. |
|
| Author: | pookpooi [ Thu Oct 14, 2010 12:01 pm ] |
| Post subject: | Re: The just over Olympiad |
Erica (19 x 19 winner) lost to a professional child even with 6 stone handicap in exhibition match. So professional players are still safe at least in a next decade (19 x 19 of course). While there's no big event in 9 x 9 human-computer match recently, I assume that these programs (especially the winner 'My Go Friend') have reach professional level, though not top it yet. |
|
| Author: | oren [ Thu Oct 14, 2010 12:31 pm ] |
| Post subject: | Re: The just over Olympiad |
pookpooi wrote: So professional players are still safe at least in a next decade (19 x 19 of course). I don't know about a decade, but at least a few years. There have been a lot of gains over the last decade. |
|
| Author: | Mike Novack [ Thu Oct 14, 2010 2:33 pm ] |
| Post subject: | Re: The just over Olympiad |
oren wrote: pookpooi wrote: So professional players are still safe at least in a next decade (19 x 19 of course). I don't know about a decade, but at least a few years. There have been a lot of gains over the last decade. Safe at what handicap? Just how many stones does this child need against a high ranked pro and then add six. Nor is it clear which of the three top programs actually the "strongest". I don't know the agreed formula used to determine the event winner, etc. but as I mentioned before this was a "round robin" tournament and the top three programs all had the SAME win/loss record. And slightly misleading to say a lot of gains over the last decade because the last big jump was a half dozen or so stones in just the last few years. Not little step by step gains in strength but a breakthrough using a completely different approach. No reason to believe another such breakthrough will occur in which case future progress in gaining strength might be slow. Or could happen tomorrow (and then a year or two for implementations to appear). |
|
| Author: | palapiku [ Thu Oct 14, 2010 2:47 pm ] |
| Post subject: | Re: The just over Olympiad |
Mike Novack wrote: Just how many stones does this child need against a high ranked pro "This child", 1p, daughter of a pro and granddaughter of Fujisawa Shuko, probably doesn't need very many. |
|
| Author: | pookpooi [ Fri Oct 15, 2010 12:17 am ] |
| Post subject: | Re: The just over Olympiad |
There's news that computer Shogi (Japanese Chess) 'Akara 2010' just beat top female shogi professional player in the even game. Computer Xiangqi (Chinese Chess) have already won over human back in 2006. What does the future hold? |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|