Who's playing in Tygem? -- It was AlphaGo!!

[Uberdude]

I started a thread to study some of Master's interesting moves: viewtopic.php?f=15&t=13929.

Also some comments on Ohashi Hirofumi 6p's blog in Japanese on Master vs Iyama Yuta: http://blog.goo.ne.jp/minamijyuujisei_1 ... ca7943cf85.

[pookpooi]

Any update on opponent today?
We've Chen Yaoye, Nie wieping, Shin Jinseo, but on which game no.?

[pookpooi]

Master said he's AlphaGo?

[Cassandra]

Master said he's AlphaGo?

Google translation of
"我是AlphaGo的黃博士"
results in
"I'm Dr. Huang of AlphaGo"

[pookpooi]

Case closed

[Laerthd]

Has anyone seen this posted from the deepmind team?
I've looked at their website and twitter ( Demis Hassabi, DeepMindAI) and cannot find anything
edit: Also, why is the sentence "I'm Dr. Huang of AlphaGo" and not "I'm Dr. Huang of Deepmind"?

[xiayun]

Any update on opponent today?
We've Chen Yaoye, Nie wieping, Shin Jinseo, but on which game no.?

Game 51 - Zhou Junxun
Game 52 - Fan Tingyu
Game 53 - Huang Yunsong
Game 54 - Nie Weiping
Game 55 - Chen Yaoye
Game 56 - Cho Hanseung
Game 57 - Shin Jinseo
Game 58 - Chang Hao
Game 59 - Zhou Ruiyang
Game 60 - Gu Li

[Bonobo]

Google translation of
"我是AlphaGo的黃博士"
results in
"I'm Dr. Huang of AlphaGo"

• Would this necessarily mean that it was AlphaGo that played? What do we know about DeepMind’s naming scheme? Following the naming traditions in software development—could BetaGo GammaGo (Gold?) “Master” be a similar (i.e. further “evolved”) engine, but perhaps totally self-trained instead of having been trained with professional games?
  
  
• I don't really know about tradition in the academic field … how common is it that a graduated person introduces themselves as “Dr. NN” rather than “Firstname Lastname”? (Yes, I know that “Dr.” is used as part of the name)
  
  
• What about the possibility of “three-brain” playing (as Ingo Althöfer calls it): a bot-assisted human player?

[xiayun]

Has anyone seen this posted from the deepmind team?
I've looked at their website and twitter ( Demis Hassabi, DeepMindAI) and cannot find anything
edit: Also, why is the sentence "I'm Dr. Huang of AlphaGo" and not "I'm Dr. Huang of Deepmind"?

Ke Jie confirmed on weibo as well: http://weibo.com/u/2865101843

Translation: "Thanks for the shock and awe the latest version of AlphaGo provided us. As someone who knew its identity from the beginning, I'd have loved to see us human winning a blitz game. If not for being in the hospital, I'd have tried one last response that I prepared for a week...slight regret, and hope Gu Li can play to the best of our human ability in this final game of AlphaGo beta test."

[JoeS1]

Well after going through many of the games. I've learned a new proverb. When your opponent builds a wall early, attack it immediately to reduce it's influence and potential. AlphaGo typically ends up making territory while doing this or it's own influence potential.

[pookpooi]

Official announcement from DeepMind
https://twitter.com/demishassabis/statu ... 3282954240

[Bonobo]

Official announcement from DeepMind
https://twitter.com/demishassabis/statu ... 3282954240

Thank you!

Embedding the image from his tweet:

[hokusai]

Here are some videos in which Master(P) won against Park Jeonghwan, Ke Jie, Iyama Yuta and Shin Jinseo:

[mistakenot]

Look what Ali Jabarin (1p, Europe) just reported a few minutes ago in his facebook

"6 pm pairing draw for Limin cup tomorrow morning, 5:58 I'm glued to Tygem watching some mysterious account beat Park Junghwan (again) after just beating Ke Jie twice in a row, oops late for the ceremony, meet ke jie who is also late from watching the game on the way to the ceremony and he is a bit shocked.. just repeating "it's too strong".
Dunno whether it's alphago or not but it's definitely some strong computer with 20-0(and counting) score against top pros, go check out Master(P) on Tygem."

Official announcement from DeepMind
https://twitter.com/demishassabis/statu ... 3282954240

Might be a good time to update the thread title to something more informative? So people don't continue to accidentally skip the thread. Maybe something like: "AlphaGo/Master(P) undefeated in 60 online blitz games vs top pros".

OP or a moderator, interested in changing the thread title?

[pookpooi]

My impression is that how AlphaGo is able to play in Japanese rule with 6.5 komi

Are they force AlphaGo? And degrading performance is acceptable since it's already way ahead of top pro.
Or they train with new data to play specifically with Japanese rule.
Or they do something, like unsupervised training, self-train from scratch, that can be flexible in rule and komi?