[Computer-go] AlphaGo won first game!
Hiroshi Yamashita
Hiroshi Yamashita
_______________________________________________
Computer-go mailing list
Compu...@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go
Michael Alford
Hiroshi Yamashita
(;GM[1]SZ[19]
PB[Lee Sedol]
PW[AlphaGo]
RE[W+R]
KM[7.5]DT[2016-03-09]TM[7200]RU[Chinese]
;B[qd];W[dd];B[pq];W[dp];B[fc];W[cf];B[ql];W[od];B[ld];W[qc]
;B[rc];W[pc];B[re];W[of];B[pg];W[og];B[ph];W[id];B[lf];W[oh]
;B[pi];W[lh];B[kh];W[ke];B[le];W[lg];B[kg];W[kf];B[ne];W[oe]
;B[jc];W[ic];B[jd];W[ie];B[je];W[jf];B[if];W[jg];B[li];W[mi]
;B[hf];W[ih];B[mb];W[gd];B[ki];W[mj];B[kk];W[ib];B[ob];W[ml]
;B[lm];W[nc];B[nb];W[kb];B[lc];W[mm];B[ln];W[kl];B[ll];W[lk]
;B[jj];W[jl];B[hj];W[hi];B[gj];W[gf];B[ii];W[jh];B[ij];W[mn]
;B[lo];W[mo];B[lp];W[mp];B[lq];W[mq];B[im];W[qo];B[fq];W[gg]
;B[cn];W[dn];B[dm];W[fp];B[gp];W[gq];B[fr];W[co];B[en];W[do]
;B[ep];W[cm];B[dl];W[lr];B[kr];W[rb];B[jb];W[ja];B[mf];W[mh]
;B[nd];W[qj];B[pj];W[qk];B[pl];W[pk];B[ok];W[rh];B[rl];W[qf]
;B[ri];W[rf];B[pf];W[qe];B[qh];W[cc];B[bn];W[bm];B[bl];W[bo]
;B[rg];W[mr];B[po];W[jr];B[kq];W[pn];B[oo];W[qp];B[on];W[pp]
;B[op];W[qq];B[or];W[pr];B[oq];W[pd];B[qr];W[rr];B[ps];W[rs]
;B[rn];W[ro];B[qn];W[so];B[cl];W[an];B[ks];W[om];B[ol];W[ci]
;B[hh];W[hg];B[dr];W[dj];B[bq];W[cq];B[cr];W[bp];B[dq];W[br]
;B[cp];W[ap];B[ek];W[fi];B[bj];W[bi];B[pb];W[qb];B[sf];W[rd]
;B[ai];W[ah];B[aj];W[bh];B[gi];W[fj];B[fk];W[oc];B[mc];W[cj]
;B[al];W[nm];B[pm];W[aq];B[gh];W[fh])
René van de Veerdonk
Jim O'Flaherty
Marc Landgraf
the game, that Q5 should have been better at R4. I would say this was
the final stage of the middle game. The result from the game left Lee
Sedol with an unwinnable endgame. And "by resignation" is meaningless
here. It is just a matter of personal preference if pros resign heir
close games, even if their are lost by 0.5 or if they decide to
resign. In this game most counts had AlphaGo 3-6 points ahead.
Tobias Pfeiffer
Wow, congrats to the AlphaGo team! Would love to see a more detailed analysis later today.
Sergey Nikolenko
commented broadcast uploaded here:
http://www.youtube.com/watch?v=vFr3K2DORc8
I don't play well enough to understand how good the commentary is, though.
With best regards,
Sergey Nikolenko.
Igor Polyakov
started playing slow moves when it got ahead. After the big fights were
finished it already knew it was going to win easily.
David Fotland
Many Faces thought alpha go was ahead most of the game. It looked to me like the turning point was when Alphago cut in the center then gave up the two cutting stones for gains on both sides (but not so strong…).
Congratulations Aja!
I watched it at Google in Mountain View with about 100 people.
David
Marc Landgraf
how does it compare to the version used against Fan Hui?
Julian Schrittwieser
As Demis said during the press conference, it's roughly the same amount of hardware.
"Ingo Althöfer"
Von: "David Fotland" <fot...@smart-games.com>
> Many Faces thought alpha go was ahead most of the game.
Similar with CrazyStone. After move 26 CS gave 56 % for AlphaGo
and never went below this value. Soon later it were 60+ %, and
never went lower, too.
Ingo.
Darren Cook
Ingo wrote:
> Similar with CrazyStone. After move 26 CS gave 56 % for AlphaGo
> and never went below this value. Soon later it were 60+ %, and
> never went lower, too.
Did it show jumps at some of the key moves the human experts thought
were decisive? (E.g. white 80, then 84-88 as poor, then either black 119
or black 123 as the losing move):
https://gogameguru.com/alphago-defeats-lee-sedol-game-1/
(I guess the "reversal" at white 136 was just an MCTS program knowing it
is ahead, and playing for probability, not winning margin.)
I.e. is it fair to say that other computer programs will appreciate and
understand the computer's moves better than the human's moves, so saying
it is ahead is to be expected? (confirmation bias)
Darren
P.S. Lee Sedol says he was shocked, and never expected to lose, even
when he was behind. I wonder if he did any special preparation for this
match? (E.g. playing handicap games against other strong MCTS program,
to appreciate how they behave when they have a lead.)
Olivier Teytaud
--
"I will never sign a document with logos in black & white." A. Einstein
Olivier Teytaud, olivier...@inria.fr, http://www.slideshare.net/teytaud
Xavier Combelle
Xavier Combelle
Hiroshi Yamashita
But it is because Aya thought Black N-18 group is 65% alive at move 75.
This Black must live 100%. So I think MCTS tend to think White is ahead.
I'm curious whether AlphaGo understands this group is over 80% alive.
Thanks,
Hiroshi Yamashita
----- Original Message -----
From: ""Ingo Althöfer"" <3-Hirn...@gmx.de>
To: <compu...@computer-go.org>
Sent: Wednesday, March 09, 2016 7:01 PM
Subject: Re: [Computer-go] AlphaGo won first game!
Richard Lorentz
His partner, on the other hand, (I forget his name) was more of a hindrance to the presentation than anything. Towards the end of the game I found him downright annoying.
-Richard
Gonçalo Mendes Ferreira
Myungwan Kim. They don't run out of magnetic stones.
https://www.youtube.com/watch?v=YZPKR7HzM_s
On 09/03/2016 16:13, Richard Lorentz wrote:
> I found Michael Redmond's commentary very good. Helped a weak player
> like me understand what was going on, but occasionally went over my
> head, which I'm sure others appreciated. He came across as a class act.
>
> His partner, on the other hand, (I forget his name) was more of a
> hindrance to the presentation than anything. Towards the end of the game
> I found him downright annoying.
>
> -Richard
>
>
>
"Ingo Althöfer"
I had the honour to watch round 1 of the match together with FJ Dickhut
(German 6-dan; player in the codecentric Challenges 2014 and 2015) and
Andreas Fecke (3-dan; Go cartoonist: "Stones").
Here are some photos with comments:
http://althofer.de/lee-sedol-alphago-round-1.html
Ingo.
Petr Baudis
On Wed, Mar 09, 2016 at 04:43:23PM +0900, Hiroshi Yamashita wrote:
> AlphaGo won 1st game against Lee Sedol!
Well, I have to eat my past words - of course, there are still four
games to go, but the first round does not look like a lucky win at all!
Huge congratulations to the AlphaGo team, you have done truly amazing
work, with potential to spearhead a lot of further advances in AI in
general! It does seem to me that you must have made a lot of progress
since the Nature paper though - is that impression correct?
Do you have some more surprising breakthroughs and techniques in store
for us, or was the progress mainly incremental, furthering the training
etc.?
By the way, there is a short snippet in the paper that maybe many
people overlooked (including me on the very first read!):
> We introduce a new technique that caches all moves from the search
> tree and then plays similar moves during rollouts; a generalisation of
> the last good reply heuristic. At every step of the tree traversal, the
> most probable action is inserted into a hash table, along with the
> 3 × 3 pattern context (colour, liberty and stone counts) around both the
> previous move and the current move. At each step of the rollout, the
> pattern context is matched against the hash table; if a match is found
> then the stored move is played with high probability.
This looks like it might overcome a lot of weaknesses re semeai etc.,
enabling the coveted (by me) information flow from tree to playouts, if
you made this to work well (it's similar to my "liberty maps" attempts,
which always failed though - I tried to encode a larger context, which
maybe wasn't good idea).
Would you say this improvement is important to AlphaGo's playing
strength (or its scaling), or merely a minor tweak?
Thanks,
--
Petr Baudis
If you have good ideas, good data and fast computers,
you can do almost anything. -- Geoffrey Hinton
Olivier Teytaud
--
=========================================================
Olivier Teytaud, INRIA TAO Research Fellow --- http://www.slideshare.net/teytaud
valk...@phmp.se
I think the technique of hashing move pairs from the search tree and
reuse them in
the playouts if the context matches, could plausibly be the major
improvement of Alphago that
we witnessed today.
Another thing I noticed is that Alphago does not use any statistics from
the playouts other than wins and losses.
I think that is a improvement because it removes biases that might make
the program weak in certain situations.
On the other hand using the powerful move prediction accuracy from the
tree search and reuse that information in
playouts could really solve a lot of problems. In the way I see it
Alphago injects go knowledge into the playouts fro mthe search tree.
Traditional monte carlo programs would do the opposite. Add a lot of
knowledge to playouts and then try to squeeze
out as much statistics as possible.
Also the clever thing is that the playouts get fed knowledge from both
offline and online sources. When the search starts
the move ordering of the neural networks will help playouts with
suggestion of good shape moves. But as more and more
local situations are read out somewhere in the search tree, the playouts
will pick up more and more strong moves from the hashtable.
By the way I will start experiments with this soon in my new program!...
:-)
Best
Magnus Persson
_______________________________________________
wing
Lee did not play any kos, except one minor irrelevant
one in the lower left. This game was so simple that
the program could accurately model the whole board.
If Lee wants to win, he needs to start 2 or 3
simultaneous kos.
Michael Wing
> Congratulations, AlphaGo and team. And by resignation! That's
> fantastic!
>
> Anyone know where the tipping point was? Did Sedol get the end game
> order just slightly off and AlphaGo took advantage? Or was their an
> earlier poor move by Sedol and/or surprising (and good) move by
> AlphaGo? I'm WAY too weak a player to even make stupid guesses. Any
> links to in depth analysis would be greatly appreciated!
>
> On Wed, Mar 9, 2016 at 1:46 AM, René van de Veerdonk
> <rene.vand...@gmail.com> wrote:
>
>> wow .. congrats to the AlphaGo team!!
>>
>> On Tue, Mar 8, 2016 at 11:43 PM, Hiroshi Yamashita
>> <y...@bd.mbn.or.jp> wrote:
>>
>>> AlphaGo won 1st game against Lee Sedol!
>>>
>>> Hiroshi Yamashita
>>>
>>> _______________________________________________
>>> Computer-go mailing list
>>> Compu...@computer-go.org
>>> http://computer-go.org/mailman/listinfo/computer-go [1]
>>
>> _______________________________________________
>> Computer-go mailing list
>> Compu...@computer-go.org
>> http://computer-go.org/mailman/listinfo/computer-go [1]
>
>
>
> Links:
> ------
> [1] http://computer-go.org/mailman/listinfo/computer-go
Hideki Kato
Pasky,
It's too early to conclude any, I think, because no records
of losing games have been published, ie., no weakpoints of
AlphaGo are open. I believe that the essential problems which
come from current MCTS (bottom-up) framework, such as solving
complex semeai's and double-ko's, aren't solved yet.
Hideki
Petr Baudis: <20160309171...@machine.or.cz>:
> Hi!
>
>
>
>
>etc.?
>
>
>
>
>
>
>
> Thanks,
>
>--
> Petr Baudis
>_______________________________________________
>Computer-go mailing list
>http://computer-go.org/mailman/listinfo/computer-go
--
Hideki Kato <mailto:hideki...@ybb.ne.jp>
David Fotland
Schaeffer and Fotland still predict Sedol will win the match. “I think the pro will win,” Fotland says, “But I think the pro will be shocked at how strong the program is.”
David Fotland
Yes, I think the programs will have similar biases. In this game Sedol had some groups that were alive, but needed correct responses to stay alive. Even though the pro's stones won’t die, the playouts sometimes manage to kill them. This makes the program think it is more ahead than it actually is. AlphaGo should be much more accurate because it has a value network and can replay sequences from the mains search.
David
>
> I.e. is it fair to say that other computer programs will appreciate and
> understand the computer's moves better than the human's moves, so saying
> it is ahead is to be expected? (confirmation bias)
>
> Darren
Petr Baudis
> I predicted Sedol would be shocked. I'm still routing for Sedol. From Scientific American interview...
>
> Schaeffer and Fotland still predict Sedol will win the match. “I think the pro will win,” Fotland says, “But I think the pro will be shocked at how strong the program is.”
In that case it's time for Lee Sedol to start working hard on turning
this match around, because AlphaGo won the second game too! :)
Petr Baudis
Petri Pitkanen
Petr Baudis
Sedol said that while he saw some questionable moves by AlphaGo in the
first game, he feels that the second game was a near-perfect play by
AlphaGo and he did not feel ahead at any point of the game.
--
Petr Baudis
If you have good ideas, good data and fast computers,
you can do almost anything. -- Geoffrey Hinton
Erik van der Werf
"Ingo Althöfer"
Von: "Erik van der Werf" <erikvan...@gmail.com>
> Very impressive results so far!
indeed, almost unbelievable.
> If it's going to be a clean sweep, I hope we will get to see some handicap games :-)
I have another proposal, IF a clean sweep will happen:
There was an announcement three days ago by a Chinese group that
they are developing a strong go bot and want to challenge
No. 1 player Ke Jie (still in 2016).
The winner of that match might challenge AlphaGo.
Ingo.
Robert Jasiek
> such as solving complex semeai's and double-ko's, aren't solved yet.
To find out Alphago's weaknesses, there can be, in particular,
- this match
- careful analysis of its games
- Alphago playing on artificial problem positions incl. complex kos,
complex ko fights, complex sekis, complex semeais, complex endgames,
multiple connection problems, complex life and death problems (such as
Igo Hatsu Yoron 120) etc., and then theoretical analysis of such play
- semantic verification of the program code and interface
- theoretical study of the used theory and the generated dynamic data
(structures)
--
robert jasiek
Jim O'Flaherty
Jim O'Flaherty
Darren Cook
> AlphaGo and see how it responded to [...complex kos, complex ko fights,
> complex sekis, complex semeais, ..., multiple connection problems, complex
> life and death problems] as ammunition for his next game.
In fact in game 2, white 172 was described [1] as the losing move,
because it would have started a ko. (I suspect the game was already
lost, and that lesser MCTS programs would have managed correct play from
there, but it would still have been useful to see.)
(By the way, in game 2, black 43 and 45 were described as "a little
heavy". It did seem (to my weak eyes) to turn out poorly. I'm curious if
this was a real mistake by AlphaGo, or if it was already happy it was
leading, and this was the one it felt led to the safest win? I really
hope the AlphaGo team will publish what it thought its winrate was after
each move.)
Darren
[1]: https://gogameguru.com/alphago-races-ahead-2-0-lee-sedol/
Darren Cook
"would have started a ko" --> "should have instead started a ko"
wing
Another question is whether he will exploit those weaknesses.
Lee has a very simple style of play that seems less ko-oriented
than other players, and this may play into the hands of Alpha.
Michael Wing
>> http://computer-go.org/mailman/listinfo/computer-go [1]
>
>
>
> Links:
> ------
> [1] http://computer-go.org/mailman/listinfo/computer-go
>
Lukas van de Wiel
Marco Scheurer
marco
Lukas van de Wiel
Robert Jasiek
> in game 2, black 43 and 45 were described as "a little
> heavy". It did seem (to my weak eyes) to turn out poorly. I'm curious if
> this was a real mistake by AlphaGo, or if it was already happy it was
> leading, and this was the one it felt led to the safest win?
In human terms, it was a combination of: limitation of the expansion
potential of the white left side, shinogi, sente and developing the
potential of the upper side including its center potential. Ugly and
marvellous strategy of simplifying the game (same: reduction of the
right side in sente) and creating a winning position by robbing White of
every option of creating significant new territory regions / expansions.
--
robert jasiek
Josef Moudrik
Yes, but they are not some random cherry picking third party; have a look on the top authors of the paper - David Silver, Aja Huang, Chris Maddison..
Regards,
Josef
Sorin Gherman
Thomas Wolf
Recent strong computer programs never loose by a few points. They are either
crashed before the end game starts (because when being clearly behind they play more
desperate and weaker moves because they mainly get negative feadback from
their search with mostly loosing branches and risky play gives them the only
winning sequences in their search) or they win by resignation or win
by a few points.
In other words, if a human player playing AlphaGo does not have a large
advantage already in the middle game, then AlphaGo will win whether it looks
like it or not (even to a 9p player like Michael Redmond was surprised
last night about the sudden gain of a number of points by AlphaGo in the
center in the end game: 4:42:10, 4:43:00, 4:43:28 in the video
https://gogameguru.com/alphago-2/)
In the middle and end game the reduced number of possible moves and the
precise and fast counting ability of computer programs are superior. In the
game commentary of the 1st game it was mentioned that Lee Sedol considers the
opening not to be his strongest part of the game. But with AlphaGo playing
top pro level even in the opening, a large advantage after the middle game
might simply be impossible to reach for a human.
About finding weakness:
In the absense of games of AlphaGo to study it might be interesting
to get a general idea by checking out the games where 7d Zen lost on KGS
recently.
Thomas
Petr Baudis
> Yes, but they are not some random cherry picking third party; have a look
> on the top authors of the paper - David Silver, Aja Huang, Chris Maddison..
Also, they aren't merely wrapping engineering around existing science
and putting existing things together, but invented several new methods
too. So, of course they are standing on the shoulders of giants, and
the massive computational resources of Google had been a lot of help,
but I'd say there is a fair amount of originality in the AlphaGo
research, scientifically.
Petr Baudis
uurtamo .
Quick question - how, mechanically, is the opening being handled by alpha go and other recent very strong programs? Giant hand-entered or game-learned joseki books?
Thanks,
steve
Sorin Gherman
From reading their article, AlphaGo makes no difference at all between start, middle and endgame.
Just like any other position, the empty (or almost empty, or almost full) board is just another game position in which it chooses (one of) the most promising moves in order to maximize her chance of winning.
Thomas Wolf
advantages of one opening move compared to another opening move is less
efficient than the learning done in the middle and end game.
uurtamo .
If that's the case, then they should be able to give opinions on best first moves, best first two move combos, and best first three move combos. That'd be interesting to see. (Top 10 or so of each).
s.
Sorin Gherman
For that reason I guess that AlphaGo opening style is mostly influenced by the net that is trained on strong human games, while as the game progresses the MC rollouts have more and more influence in choosing a move.
Is my understanding way off?
Olivier Teytaud
Olivier Teytaud, olivier...@inria.fr, TAO, LRI, UMR 8623(CNRS - Univ. Paris-Sud),
bat 490 Univ. Paris-Sud F-91405 Orsay Cedex France http://www.slideshare.net/teytaud
Thomas Wolf
games, there are at least in the opening many moves with the same optimal
outcome. The difference between these moves is not the guaranteed score (they
are all optimal) but the difficulty to play optimal after that move. And the
human and computer strengths are rather different.
Jim O'Flaherty
Brian Sheppard
Amen to Don Dailey. He would be so proud.
terry mcintyre
Sent from Yahoo Mail for iPad
Seo Sanghyeon
> Hypothetically, they could have grafted one on. I read a report that the
> first move in game 2 vs. Lee Sedol took only seconds. On the other hand,
> it's first move in game 1 took a longer while. We can only speculate.
This is easy to explain. AlphaGo was white (second to play) in game 1,
and black (first to play) in game 2. You can precalculate a move if you are
first to play. Harder to do that if you are second.
--
Seo Sanghyeon
uurtamo .
Not to put too fine a point on it, but there's not very many two or three-move combos on an empty board. As staggering as it is, I'm inclined to believe without further evidence that there's no book or just a very light book.
s.
David Fotland
David
> -----Original Message-----
> From: Computer-go [mailto:computer-...@computer-go.org] On Behalf
> Of Darren Cook
> Sent: Thursday, March 10, 2016 8:26 AM
> To: compu...@computer-go.org
> Subject: Re: [Computer-go] Finding Alphago's Weaknesses
>