This document discusses using Meta-Monte Carlo Tree Search (Meta-MCTS) to build an opening book for 7x7 Go. Meta-MCTS improved its play against a sparring partner that incorporated human variations. While Meta-MCTS won all games as black and white against professionals, humans found at least one variation where it did not play correctly. The document concludes that Meta-MCTS performed well but incorporating human data helped, and exactly solving 7x7 Go would require immense work collecting and solving all leaf variations.
2. - Go is not solved beyond 6x6
- We build an opening book for 7x7 for
approximately solving 7x7 Go..
Widely believed: perfect play is a draw with Komi 9.
3. Our tools
1) Monte-Carlo Tree Search
2) Meta-Monte-Carlo Tree Search
3) Senseis' partial solution
(using in particular Davies' work)
23. Meta-MCTS learns against a
MCTS sparring partner.
We introduce Senseis' variations
into this sparring partner
during the Meta-MCTS run.
24. Learning curve of black by Meta-MCTS:
- X-axis = log2(number of playouts)
- Y-axis = moving average (window size 55)
of winning rate in playouts
Playouts = MCTS (it's Meta-MCTS)
Komi = 8.5 (winning with komi 8.5
ensures a draw with komi 9)
25. Learning curve of white by Meta-MCTS:
- X-axis = log2(number of playouts)
- Y-axis = moving average (window size 55)
of winning rate in playouts
Playouts = MCTS (it's Meta-MCTS)
Komi = 9.5 (winning with komi 9.5
ensures a draw with komi 9)
26. Decreasing points in the curve = introduction
of Senseis variations in the opponent.
Conclusion = the algorithm did not find alone
all these variations ==> human needed.
30. With komi 9.5, MoGoTW won everything as White.
With komi 8.5, MoGoTW won everything as Black.
Exciting!
Were all MoGoTW's moves perfect ?
31. With komi 9.5, MoGoTW won everything as White.
With komi 8.5, MoGoTW won everything as Black.
Exciting!
Were all MoGoTW's moves perfect ?
No :-(
In one game (at least) the human might
have won.
32. Left: this game was won by MoGoTW as black.
Chun-Yen Lin (2P) made a mistake.
Right: how Chun-Yen Lin (2P) might have won the game.
33. So, still at least one variation on
which the bot does not play correctly.
We did not introduce manually a correction,
but we introduced the variation played by the pro
in the sparring partner.
We see if the bot can find a solution by itself.
34. Learning curve as black, after introducing the
dangerous variation in the sparring partner.
Time still logarithmic.
35. CONCLUSIONS
We used Meta-MCTS for building
an opening book for 7x7 Go.
We are not aware of
remaining bad moves, which does
not mean there's no more bad move.
36. Meta-MCTS did a good job by itself,
but human inputs
( = Senseis + games against
pros) have been helpful.
Towards exact solving ?
= collecting all leafs of the OB
+ solving all of them...
= huge work.
37. Other conclusions:
- 7x7 can be very hard, even for pros
(pros made mistakes).
- MCTS alone is not enough for
very strong play in 7x7