Difference between revisions of "MCαβ"
m |
m |
||
Line 2: | Line 2: | ||
'''Monte-Carlo Alpha-Beta''', '''(MCαβ)'''<br/> | '''Monte-Carlo Alpha-Beta''', '''(MCαβ)'''<br/> | ||
− | is a [[Best-First|Best-First search]] algorithm based on [[Monte-Carlo Tree Search]] and a shallow [[Alpha-Beta|alpha-beta]] [[Depth-First|depth-first-search]]. It is used in [[Lines of Action]] and in conjunction with [[UCT]] (Upper Confidence bounds applied to Trees) also in Chess. It is similar to the the [[Best-First Minimax Search|Best-first minimax search]] proposed by [[Richard Korf]] and [[Max Chickering]] <ref>[[Richard Korf]], [[Max Chickering]] ('''1996'''). ''[https://www.microsoft.com/en-us/research/publication/best-first-minimax-search/ Best-First Minimax Search]''. [https://en.wikipedia.org/wiki/Artificial_Intelligence_(journal) Artificial Intelligence], Vol. 84, No 1-2</ref>, one distinction between the two might be that MCTS uses usually win/draw/loss scores and | + | is a [[Best-First|Best-First search]] algorithm based on [[Monte-Carlo Tree Search]] and a shallow [[Alpha-Beta|alpha-beta]] [[Depth-First|depth-first-search]]. It is used in [[Lines of Action]] and in conjunction with [[UCT]] (Upper Confidence bounds applied to Trees) also in Chess. It is similar to the the [[Best-First Minimax Search|Best-first minimax search]] proposed by [[Richard Korf]] and [[Max Chickering]] <ref>[[Richard Korf]], [[Max Chickering]] ('''1996'''). ''[https://www.microsoft.com/en-us/research/publication/best-first-minimax-search/ Best-First Minimax Search]''. [https://en.wikipedia.org/wiki/Artificial_Intelligence_(journal) Artificial Intelligence], Vol. 84, No 1-2</ref>, one distinction between the two might be that MCTS uses usually win/draw/loss scores and Best-First a score from a heuristic evaluation function. |
<span id="Four Phases"></span> | <span id="Four Phases"></span> | ||
=Four Phases= | =Four Phases= |
Revision as of 10:41, 8 October 2022
Monte-Carlo Alpha-Beta, (MCαβ)
is a Best-First search algorithm based on Monte-Carlo Tree Search and a shallow alpha-beta depth-first-search. It is used in Lines of Action and in conjunction with UCT (Upper Confidence bounds applied to Trees) also in Chess. It is similar to the the Best-first minimax search proposed by Richard Korf and Max Chickering [1], one distinction between the two might be that MCTS uses usually win/draw/loss scores and Best-First a score from a heuristic evaluation function.
Four Phases
MCαβ can be divided into four strategic phases, repeated as long as there is time left:
- In the Selection phase the tree is traversed from the root node until it selects a leaf node that is not added to the tree yet
- The Expansion strategy adds the leaf node to the tree
- The Playout phase performs a shallow alpha-beta search
- The Backpropagation strategy propagates the results through the tree
See also
- Alpha-Beta
- Best-First Minimax Search
- Minimax
- Monte-Carlo Tree Search
- MT-SSS*
- Rocinante
- Rollout Paradigm
Publications
- Richard Korf, Max Chickering (1996). Best-First Minimax Search. Artificial Intelligence, Vol. 84, No 1-2
- Pim Nijssen, Mark Winands (2011). Playout Search for Monte-Carlo Tree Search in Multi-Player Games. Advances in Computer Games 13, pdf
- Cameron Browne, Edward Powley, Daniel Whitehouse, Simon Lucas, Peter Cowling, Philipp Rohlfshagen, Stephen Tavener, Diego Perez, Spyridon Samothrakis, Simon Colton (2012). A Survey of Monte Carlo Tree Search Methods. IEEE Transactions on Computational Intelligence and AI in Games, Vol. 4, No. 1, pdf
- Hendrik Baier, Mark Winands (2013). Monte-Carlo Tree Search and minimax hybrids. CIG 2013, pdf
- Hendrik Baier, Mark Winands (2014). Monte-Carlo Tree Search and Minimax Hybrids with Heuristic Evaluation Functions. ECAI CGW 2014
- Bojun Huang (2015). Pruning Game Tree by Rollouts. AAAI » MCTS, MT-SSS*, Rollout Paradigm [2]
- Hendrik Baier (2017). A Rollout-Based Search Algorithm Unifying MCTS and Alpha-Beta. Computer Games
Forum Posts
- MCTS without random playout? by Antonio Torrecillas, CCC, January 01, 2012 » B*
- Help with Best-First Select-Formula by Srdja Matovic, CCC, July 23, 2012
- Re: Announcing lczero by Daniel Shawul, CCC, January 21, 2018 » Leela Chess Zero
- Alpha-Beta as a rollouts algorithm by Daniel Shawul, CCC, January 25, 2018 » Alpha-Beta, Monte-Carlo Tree Search, Scorpio
External Links
- Larry Coryell & The Eleventh House, Oslo 1975, YouTube Video
- feat. Mike Lawrence, Alphonse Mouzon, John Lee, Mike Mandel
References
- ↑ Richard Korf, Max Chickering (1996). Best-First Minimax Search. Artificial Intelligence, Vol. 84, No 1-2
- ↑ Re: Announcing lczero by Daniel Shawul, CCC, January 21, 2018 » Leela Chess Zero