A Survey of Monte Carlo Tree Search Methods
The DOI matches its Crossref record; 220 of the 233 references it cites were found in a registry or an index.
Abstract
Monte Carlo tree search (MCTS) is a recently proposed search method that combines the precision of tree search with the generality of random sampling. It has received considerable interest due to its spectacular success in the difficult problem of computer Go, but has also proved beneficial in a range of other domains. This paper is a survey of the literature to date, intended to provide a snapshot of the state of the art after the first five years of MCTS research. We outline the core algorithm's derivation, impart some structure on the many variations and enhancements that have been proposed, and summarize the results from the key game and nongame domains to which MCTS methods have been applied. A number of open research questions indicate that the field is ripe for future work.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“RL algorithms such as Deep Q-Network (DQN) [9], Proximal Policy Optimization (PPO) [10], and Monte Carlo Tree Search (MCTS) [11] disregard the previous trajectories.”
From Playtesting: What is Beyond Personas (2023) · cites this paper as [11]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Artificial intelligence: a modern approach
(1995) · Choice Reviews Online · doi:10.5860/choice.33-1577 · Cited by 21,974
Artificial intelligence: A modern approach
Nilsson, N. J. (1996) · Artificial Intelligence · doi:10.1016/0004-3702(96)00007-0 · Cited by 10,583
Finite-time Analysis of the Multiarmed Bandit Problem
Auer, P., Cesa-Bianchi, N., & Fischer, P. (2002) · Machine Learning · doi:10.1023/a:1013689704352 · Cited by 6,033
Rapidly-exploring random trees : a new tool for path planning
LaValle, S. M. (1998) · Cited by 3,704
Bandit Based Monte-Carlo Planning
Kocsis, L., & Szepesvári, C. (2006) · Lecture Notes in Computer Science · doi:10.1007/11871842_29 · Cited by 2,855 · cited 10×
“We emphasise the importance of game theory, as this is the domain to which MCTS is most applied.”
BackgroundMethodInfluentialAsymptotically efficient adaptive allocation rules
Lai, T. L., & Robbins, H. (1985) · Advances in Applied Mathematics · doi:10.1016/0196-8858(85)90002-8 · Cited by 2,406
Efficient Selectivity and Backup Operators in Monte-Carlo Tree Search
Coulom, R. (2007) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-75538-8_7 · Cited by 1,240 · cited 3×
“Real-world games typically involve a delayed reward structure in which only those rewards achieved in the terminal states of the game accurately describe how well each player is performing.”
BackgroundGambling in a rigged casino: The adversarial multi-armed bandit problem
Auer, P., Cesa-Bianchi, N., Freund, Y., & Schapire, R. E. (2002) · Proceedings of IEEE 36th Annual Foundations of Computer Science · doi:10.1109/sfcs.1995.492488 · Cited by 742 · cited 1×
“[14] and further analyzed by Audibert and Bubeck [11], applies in the stochastic case (and hence also in the adversarial case).”
BackgroundMinimax policies for adversarial and stochastic bandits
Audibert, J.-Y., Est, U. P., & Bubeck, S. (2009) · HAL · Cited by 555
Sample mean based index policies by O(log n) regret for the multi-armed bandit problem
Agrawal, R. (1995) · Advances in Applied Probability · doi:10.2307/1427934 · Cited by 540
Bubeck, S., Munos, R., Stoltz, G., & Szepesvári, C. (2011) · Journal of Machine Learning Research · Cited by 515
Combining online and offline knowledge in UCT
Gelly, S., & Silver, D. (2007) · Proceedings of the 24th international conference on Machine learning · doi:10.1145/1273496.1273531 · Cited by 465 · cited 4×
“However, finding the set of agents with the correct properties (i.e. those that increase playing strength) is computationally intensive.”
BackgroundMethodInfluentialPROGRESSIVE STRATEGIES FOR MONTE-CARLO TREE SEARCH
CHASLOT, G. M. J.-B., WINANDS, M. H. M., HERIK, H. J. V. D., UITERWIJK, J. W. H. M., & BOUZY, B. (2008) · New Mathematics and Natural Computation · doi:10.1142/s1793005708001094 · Cited by 435
Monte-Carlo tree search and rapid action value estimation in computer Go
Gelly, S., & Silver, D. (2011) · Artificial Intelligence · doi:10.1016/j.artint.2011.03.007 · Cited by 367 · cited 5×
“In addition, the use of transpositions allows information to be shared between these extra nodes where the state is unchanged.”
InfluentialMonte-Carlo Tree Search: A New Framework for Game AI
Chaslot, G., Bakkes, S., Szita, I., & Spronck, P. (2008) · Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · doi:10.1609/aiide.v4i1.18700 · Cited by 312 · cited 2×
“6The simulation and expansion steps are often described and/or implemented in the reverse order in practice [52], [67].”
BackgroundCOMPUTING “ELO RATINGS” OF MOVE PATTERNS IN THE GAME OF GO1
Coulom, R. (2007) · ICGA Journal · doi:10.3233/icg-2007-30403 · Cited by 298 · cited 4×
“In Sparse UCT, a node may have several children corresponding to the same move, each child corresponding to a different stochastic outcome of that move.”
BackgroundMethodInfluentialModification of UCT with Patterns in Monte-Carlo Go
Gelly, S., Wang, Y., Munos, R., & Teytaud, O. (2006) · HAL · Cited by 290
Continuous Upper Confidence Trees
Couëtoux, A., Hoock, J., Sokolovska, N., Teytaud, O., & Bonnard, N. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-25566-3_32 · Cited by 256 · cited 2×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundInfluentialParallel Monte-Carlo Tree Search
Chaslot, G. M. J. B., Winands, M. H. M., & van den Herik, H. J. (2008) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-87608-3_6 · Cited by 237 · cited 7×
“1) Leaf Parallelization: Leaf parallelization as defined in [59] involves performing multiple simultaneous simulations every time the MCTS tree policy reaches (or creates) a leaf”
BackgroundInfluentialSimulation-based approach to general game playing
Finnsson, H., & Björnsson, Y. (2008) · Cited by 235
The history heuristic and alpha-beta search enhancements in practice
Schaeffer, J. (1989) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.42858 · Cited by 227 · cited 1×
“However, finding the set of agents with the correct properties (i.e. those that increase playing strength) is computationally intensive.”
BackgroundCazenave, T. (2009) · Base Institutionnelle de Recherche de l'université Paris-Dauphine (BIRD) (University Paris-Dauphine) · Cited by 207
Exploration exploitation in Go: UCT for Monte-Carlo Go
Gelly, S., & Wang, Y. (2006) · HAL · Cited by 190
Pure exploration in finitely-armed and continuous-armed bandits
Bubeck, S., Munos, R., & Stoltz, G. (2011) · Theoretical Computer Science · doi:10.1016/j.tcs.2010.12.059 · Cited by 188
CadiaPlayer: A Simulation-Based General Game Player
Björnsson, Y., & Finnsson, H. (2009) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2009.2018702 · Cited by 184 · cited 2×
“This section describes some of the characteristics that make MCTS a popular choice of algorithm for a variety of domains, often with notable success.”
BackgroundMethodInfluentialMonte Carlo Sampling for Regret Minimization in Extensive Games
Bowling, M., Zinkevich, M., Waugh, K., & Lanctot, M. (2009) · doi:10.7939/r3319s48q · Cited by 182 · cited 2×
“Coquelin and Munos [68] demonstrate that flat UCB retains the adaptivity of standard UCT while improving its regret bounds in certain worst cases where UCT is overly optimistic.”
BackgroundInfluentialFuego—An Open-Source Framework for Board Games and Go Engine Based on Monte Carlo Tree Search
Enzenberger, M., Muller, M., Arneson, B., & Segal, R. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2083662 · Cited by 174 · cited 1×
“In addition, the use of transpositions allows information to be shared between these extra nodes where the state is unchanged.”
BackgroundBandit Algorithms for Tree Search
Coquelin, P.-A., & Munos, R. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1408.2028 · Cited by 170
A Monte-Carlo AIXI Approximation
Veness, J., Ng, K. S., Hutter, M., Uther, W., & Silver, D. (2011) · Journal of Artificial Intelligence Research · doi:10.1613/jair.3125 · Cited by 164
Allis, L. V., van der Meulen, M., & van den Herik, H. J. (1994) · Artificial Intelligence · doi:10.1016/0004-3702(94)90004-3 · Cited by 162 · cited 1×
“Proof-number search (PNS) is a standard AI technique for proving game-theoretic values, typically used for endgame solvers, in which terminal states are considered to be proven wins or losses and deductions chained backwards from these [4].”
UCT for tactical assault planning in real-time strategy games
Balla, R.-K., & Fern, A. (2009) · Cited by 158
Kocsis, L., Szepesvári, C., & Willemson, J. (2006) · Cited by 158
The Computational Intelligence of MoGo Revealed in Taiwan's Computer Go Tournaments
Lee, C.-S., Wang, M., Chaslot, G. M. J. B., Hoock, J., Rimmel, A., Teytaud, O., Tsai, S.-R., Hsu, S.-C., & Hong, T. (2009) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2009.2018703 · Cited by 158 · cited 3×
“Borsboom et al. [23] suggest ways of combining UCT with HOP-like ideas, namely early probabilistic guessing and late random guessing .”
BackgroundMethodGIB: Imperfect Information in a Computationally Challenging Game
Ginsberg, M. L. (2001) · Journal of Artificial Intelligence Research · doi:10.1613/jair.820 · Cited by 157 · cited 1×
“The paper concludes with two tables that summarise the many variations and enhancements of MCTS and the domains to which they have been applied.”
BackgroundMonte Carlo Tree Search in Hex
Arneson, B., Hayward, R., & Henderson, P. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2067212 · Cited by 144 · cited 3×
“Marcolino and Matsubara [139] argue that the emergent properties of interactions between different agent types lead to increased exploration of the search space.”
BackgroundMethodSingle-Player Monte-Carlo Tree Search
Schadd, M. P. D., Winands, M. H. M., van den Herik, H. J., Chaslot, G. M. J. B., & UITERWIJK, J. W. H. M. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-87608-3_1 · Cited by 140 · cited 2×
“This section describes some of the characteristics that make MCTS a popular choice of algorithm for a variety of domains, often with notable success.”
BackgroundMethodOnline Optimization in X-Armed Bandits
Bubeck, S., Stoltz, G., Szepesvári, C., & Munos, R. (2008) · Cited by 132
Expected-outcome: a general model of static evaluation
Abramson, B. (1990) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.44404 · Cited by 130 · cited 1×
“In Section 3, the MCTS algorithm and its key components are described in detail.”
MethodModifications of UCT and sequence-like simulations for Monte-Carlo Go
Wang, Y., & Gelly, S. (2007) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2007.368095 · Cited by 130 · cited 2×
“Another modification used in [60] and [232] was to wait until a node had been visited a fixed number of times before calculating H i .”
MethodSheppard, B. (2002) · Artificial Intelligence · doi:10.1016/s0004-3702(01)00166-7 · Cited by 124 · cited 1×
“The paper concludes with two tables that summarise the many variations and enhancements of MCTS and the domains to which they have been applied.”
BackgroundRubin, J., & Watson, I. D. (2011) · Artificial Intelligence · doi:10.1016/j.artint.2010.12.005 · Cited by 122 · cited 2×
“Coquelin and Munos [68] demonstrate that flat UCB retains the adaptivity of standard UCT while improving its regret bounds in certain worst cases where UCT is overly optimistic.”
BackgroundInfluentialFeature Selection as a One-Player Game
Gaudel, R., & Sébag, M. (2010) · HAL · Cited by 121
Achieving master level play in 9×9 computer go
Gelly, S., & Silver, D. (2008) · UCL Discovery (University College London) · Cited by 121
Chaslot, G. M. J. B. (2010) · Research Publications (Maastricht University) · doi:10.26481/dis.20100930gc · Cited by 110
Monte-Carlo simulation balancing
Silver, D., & Tesauro, G. J. (2009) · International Conference on Machine Learning · doi:10.1145/1553374.1553495 · Cited by 109 · cited 3×
“NRPA has achieved superior results in puzzle optimisation tasks, including beating the human world record for Morpion Solitaire (7.4).”
BackgroundInfluentialLorentz, R. J. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-87608-3_2 · Cited by 107 · cited 1×
“Another approach called dynamic exploration , proposed by Bourki et al. [25], tunes parameters based on patterns in their Go program M O G O .”
BackgroundSample-based learning and search with permanent and transient memories
Silver, D., Sutton, R. S., & Müller, M. (2008) · Proceedings of the 25th international conference on Machine learning - ICML '08 · doi:10.1145/1390156.1390278 · Cited by 107 · cited 2×
“This opening book improved the playing strength of the program and was reported to be consistent with expert Go knowledge in some cases.”
BackgroundMethodIntegrating Sample-Based Planning and Model-Based Reinforcement Learning
Walsh, T., Goschin, S., & Littman, M. (2010) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v24i1.7689 · Cited by 104 · cited 1×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundInfluentialMonte-Carlo Tree Search Solver
Winands, M. H. M., Björnsson, Y., & Saito, J.-T. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-87608-3_3 · Cited by 93 · cited 2×
“Section 4.8.9 discusses this in more detail.”
BackgroundInfluentialCazenave, T., & Jouandeau, N. (2007) · Cited by 93
Automatic generation and evaluation of recombination games
Browne, C. (2008) · Cited by 92
Self-citationMonte-Carlo Tree Search in Settlers of Catan
Szita, I., Chaslot, G. M. J. B., & Spronck, P. H. M. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_3 · Cited by 91 · cited 1×
“[209] implemented a multiplayer MCTS player (Section IV-E) for Settlers of Catan, using domain knowledge based on players’ resources and current position to bias move selection.”
MethodSample-Based Planning for Continuous Action Markov Decision Processes
Mansley, C. R., Weinstein, A., & Littman, M. L. (2011) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v21i1.13484 · Cited by 91 · cited 2×
“TDL can learn heuristic value functions to inform the tree policy or the simulation (playout) policy.”
BackgroundParallel Nested Monte-Carlo search
Cazenave, T., & Jouandeau, N. (2009) · IEEE International Parallel and Distributed Processing Symposium (IPDPS) · doi:10.1109/ipdps.2009.5161122 · Cited by 90 · cited 2×
“If m = 0 then only actions selected in the tree are used to update nodes, similarly if m is larger than the number of moves in the simulation, this is equivalent to the AMAF algorithm.”
MethodInfluentialAn Analysis of UCT in Multi-player Games
Sturtevant, N. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-87608-3_4 · Cited by 87 · cited 2×
“Sturtevant [207] shows that this variant of UCT converges to an”
BackgroundDouble Updating Online Learning
Zhao, P., Hoi, S. C. H., & Jin, R. (2011) · doi:10.5555/1953048.2021051 · Cited by 87
Continuous Lunches Are Free Plus the Design of Optimal Optimization Algorithms
Auger, A., & Teytaud, O. (2008) · Algorithmica · doi:10.1007/s00453-008-9244-5 · Cited by 84 · cited 1×
“This is in contrast to plain UCT, which expands all branches indefinitely.”
ResultCombining UCT and Nested Monte Carlo Search for Single-Player General Game Playing
Méhat, J., & Cazenave, T. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2088123 · Cited by 83 · cited 2×
“This would be detrimental in games of more than one player since such a strong line would probably rely on the unrealistic assumption that the opponent plays weak moves, but for single-player games this is not an issue.”
BackgroundMethodInfluentialUnderstanding the Success of Perfect Information Monte Carlo Sampling in Game Tree Search
Long, J., Sturtevant, N., Buro, M., & Furtak, T. (2010) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v24i1.7562 · Cited by 81 · cited 2×
“Coquelin and Munos [68] demonstrate that flat UCB retains the adaptivity of standard UCT while improving its regret bounds in certain worst cases where UCT is overly optimistic.”
BackgroundLower Bounding Klondike Solitaire with Monte-Carlo Planning
Bjarnason, R., Fern, A., & Tadepalli, P. V. (2009) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v19i1.13363 · Cited by 81 · cited 2×
“Coquelin and Munos [68] demonstrate that flat UCB retains the adaptivity of standard UCT while improving its regret bounds in certain worst cases where UCT is overly optimistic.”
BackgroundInfluentialLearning Simulation Control in General Game-Playing Agents
Finnsson, H., & Björnsson, Y. (2010) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v24i1.7651 · Cited by 80 · cited 2×
“These have typically had success with single-player puzzles and similar optimisation tasks.”
BackgroundMethodInfluentialMonte-Carlo exploration for deterministic planning
Nakhost, H., & Müller, M. (2009) · Cited by 79
Transpositions and move groups in Monte Carlo tree search
Childs, B. E., Brodeur, J. H., & Kocsis, L. (2008) · 2008 IEEE Symposium On Computational Intelligence and Games · doi:10.1109/cig.2008.5035667 · Cited by 78 · cited 8×
“Randomness may arise through the shuffling of a deck of cards or the rolling of dice.”
BackgroundMethodInfluentialCurrent Frontiers in Computer Go
Rimmel, A., Teytaud, O., Lee, C.-S., Yen, S.-J., Wang, M., & Tsai, S.-R. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2098876 · Cited by 75 · cited 1×
“Opening books 16 have been used to improve playing strength in artificial players for many games.”
Monte-Carlo Tree Search in Poker Using Expected Reward Distributions
den Broeck, G. V., Driessens, K., & Ramon, J. (2009) · Lecture notes in computer science · doi:10.1007/978-3-642-05224-8_28 · Cited by 71 · cited 2×
“apply MCTS methods to multiplayer no-limit Texas Hold’em Poker [223], enabling strong exploitative behaviour against weaker rule-based opponents and competitive performance against experienced human opponents.”
BackgroundMethodMonte Carlo Tree Search in Lines of Action
Winands, M. H. M., Björnsson, Y., & Saito, J.-T. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2061050 · Cited by 71 · cited 2×
“Section 4.8.9 discusses this in more detail.”
BackgroundInfluentialBandit-based optimization on graphs with application to library performance tuning
de Mesmay, F., Rimmel, A., Voronenko, Y., & Püschel, M. (2009) · International Conference on Machine Learning · doi:10.1145/1553374.1553468 · Cited by 69 · cited 1×
“2) Threshold Ascent for Graphs (TAG): Threshold ascent for graphs (TAG) is a method that extends the MCTS paradigm by maximizing an objective function over the sinks of directed acyclic graphs [166], [73].”
MethodMonte Carlo tree search in Kriegspiel
Ciancarini, P., & Favini, G. P. (2010) · Artificial Intelligence · doi:10.1016/j.artint.2010.04.017 · Cited by 68 · cited 1×
“This allows game-theoretic values to be proven twice as quickly in computer Go experiments [183], with a quarter of the nodes.”
BackgroundGuiding Combinatorial Optimization with UCT
Sabharwal, A., Samulowitz, H., & Reddy, C. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-29828-8_23 · Cited by 65 · cited 2×
“Mixed Integer Programming In the study performed by Sabharwal and Samulowitz [179], UCT is applied to guide Mixed Integer Programming (MIP), comparing the performance of the UCT based node selection with that of CPLEX, a traditional MIP solver, and best-first, breadth-first and depth-first strategies, showing very pro…”
MethodA Lock-Free Multithreaded Monte-Carlo Tree Search Algorithm
Enzenberger, M., & Müller, M. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_2 · Cited by 61 · cited 1×
“The EXP3 policy operates as follows: • Draw an arm I t from the probability distribution p t .”
BackgroundAdding Expert Knowledge and Exploration in Monte-Carlo Tree Search
Chaslot, G. M. J. B., Fiter, C., Hoock, J., Rimmel, A., & Teytaud, O. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_1 · Cited by 61 · cited 6×
“in [53] and [55], designed to increase simulation diversity for the”
BackgroundMethodInfluentialCreating an Upper-Confidence-Tree Program for Havannah
Teytaud, F., & Teytaud, O. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_7 · Cited by 60 · cited 2×
“Borsboom et al. [23] suggest ways of combining UCT with HOP-like ideas, namely early probabilistic guessing and late random guessing .”
BackgroundMethodThe Power of Forgetting: Improving the Last-Good-Reply Policy in Monte Carlo Go
Baier, H., & Drake, P. D. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2100396 · Cited by 56 · cited 4×
“8) to inform simulations, modified by Baier and Drake [17] to include the forgetting of bad moves.”
BackgroundMethodInfluentialImproving state evaluation, inference, and search in trick-based card games
Buro, M., Long, J. R., Furtak, T., & Sturtevant, N. (2009) · Cited by 55
A Parallel General Game Player
Méhat, J., & Cazenave, T. (2011) · Künstliche Intell. · doi:10.1007/s13218-010-0083-6 · Cited by 55 · cited 1×
“The RAVE-max update rule and its stochastic variant δ -RAVE-max were found to improve performance in degenerate cases for the Sum of Switches game (7.3) but were less successful for Go.”
MethodInfluentialDeterminization and information set Monte Carlo Tree Search for the card game Dou Di Zhu
Whitehouse, D., Powley, E. J., & Cowling, P. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6031993 · Cited by 55 · cited 1×
“Coquelin and Munos [68] demonstrate that flat UCB retains the adaptivity of standard UCT while improving its regret bounds in certain worst cases where UCT is overly optimistic.”
BackgroundSelf-citationCROSS-ENTROPY FOR MONTE-CARLO TREE SEARCH
Chaslot, G. M. J. B., Winands, M. H. M., Szita, I., & van den Herik, H. J. (2008) · ICGA Journal · doi:10.3233/icg-2008-31303 · Cited by 54 · cited 1×
“used the cross-entropy method to fine tune parameters for the Go playing program MANGO [58].”
MethodMonte Carlo search applied to card selection in Magic: The Gathering
Ward, C. D., & Cowling, P. (2009) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2009.5286501 · Cited by 54 · cited 1×
“Saito et al. [183] introduce Monte Carlo proof-number search (MC-PNS), a variant of proof-number search in which nodes that do not immediately prove a game-theoretic value are evaluated by Monte Carlo simulation.”
BackgroundSelf-citationMonte-Carlo Simulation Balancing in Practice
Huang, S.-C., Coulom, R., & Lin, S. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_8 · Cited by 52 · cited 1×
“When a node has been visited only a few times and its statistics are not reliable, then more accurate information can come from a heuristic value H i for a node with index i from the current position.”
BackgroundOptimization of the Nested Monte-Carlo Algorithm on the Traveling Salesman Problem with Time Windows
Rimmel, A., Teytaud, F., & Cazenave, T. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-20520-0_51 · Cited by 52 · cited 1×
“This is related to P-game trees (7.3).”
InfluentialEnhancements for Multi-Player Monte-Carlo Tree Search
Nijssen, J. A. M., & Winands, M. H. M. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_22 · Cited by 51 · cited 1×
“Nijssen and Winands [155] describe the Multi-Player Monte-Carlo Tree Search Solver (MP-MCTS-Solver) version of their MCTS Solver enhancement (5.”
BackgroundMonte-Carlo tree search in production management problems
Chaslot, G. M. J. B., de Jong, S., Saito, J.-T., & UITERWIJK, J. W. H. M. (2006) · Research Publications (Maastricht University) · Cited by 49
On Adversarial Search Spaces and Sampling-Based Planning
Ramanujan, R., Sabharwal, A., & Selman, B. (2010) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v20i1.13437 · Cited by 48 · cited 2×
“Permutation AMAF would also update these nodes.”
BackgroundCazenave, T. (2006) · Lecture notes in computer science · doi:10.1007/11922155_9 · Cited by 47 · cited 2×
“Coquelin and Munos [68] demonstrate that flat UCB retains the adaptivity of standard UCT while improving its regret bounds in certain worst cases where UCT is overly optimistic.”
BackgroundA simple tree search method for playing Ms. Pac-Man
Robles, D., & Lucas, S. M. (2009) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2009.5286469 · Cited by 47 · cited 1×
“• Combined heuristics can be more powerful than individual heuristics.”
BackgroundSelf-citationFast Approximate Max-n Monte Carlo Tree Search for Ms Pac-Man
Samothrakis, S., Robles, D., & Lucas, S. M. (2011) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2011.2144597 · Cited by 47
Self-citationReinforcement Learning and Simulation-Based Search in Computer Go
Silver, D. (2009) · doi:10.7939/r39d8t · Cited by 46 · cited 2×
“This section describes some of the characteristics that make MCTS a popular choice of algorithm for a variety of domains, often with notable success.”
BackgroundBiasing Monte-Carlo Simulations through RAVE Values
Rimmel, A., Teytaud, F., & Teytaud, O. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_6 · Cited by 44 · cited 1×
“Databases of opening move sequences of known utility.”
BackgroundOn the Scalability of Parallel UCT
Segal, R. B. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_4 · Cited by 43 · cited 1×
“EXP3 has been used in conjunction with UCT to address games with partial observability and simultaneous moves [216], [217].”
MethodScalability and Parallelization of Monte-Carlo Tree Search
Bourki, A., Chaslot, G. M. J. B., Coulm, M., Danjean, V., Doghmen, H., Hoock, J., Thomas Herault, A., Rimmel, A., Teytaud, F., Teytaud, O., et al. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_5 · Cited by 43 · cited 2×
“AIXI is a mathematical approach based on a Bayesian optimality notion for general reinforcement learning agents.”
BackgroundMethodA Parallel Monte-Carlo Tree Search Algorithm
Cazenave, T., & Jouandeau, N. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-87608-3_7 · Cited by 43 · cited 1×
“The Exploration-Exploitation with Exponential weights (EXP3) algorithm, originally proposed by Auer et al. [14] and further analysed by Audibert and Bubeck [11], applies in the stochastic case (and hence also in the adversarial case).”
MethodMonte-Carlo tree search in Ms. Pac-Man
望, 池., & Ito, T. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6031987 · Cited by 43 · cited 1×
“• Combined heuristics can be more powerful than individual heuristics.”
BackgroundAll-Moves-As-First Heuristics in Monte-Carlo Go
Helmbold, D. P., & Parker-Wood, A. (2009) · Cited by 42
Bayesian Inference in Monte-Carlo Tree SearchPreprint
Tesauro, G. J., Rajan, V. T., & Segal, R. B. (2012) · arXiv (Cornell University) · doi:10.48550/arxiv.1203.3519 · Cited by 41 · cited 1×
“[213] propose that the Bayesian framework potentially allows much more accurate”
BackgroundIntelligent Agents for the Game of Go
Hoock, J., Lee, C.-S., Rimmel, A., Teytaud, F., Wang, M., & Teytaud, O. (2010) · IEEE Computational Intelligence Magazine · doi:10.1109/mci.2010.938360 · Cited by 40 · cited 2×
“In order to make the complexity of real-time video games tractable, approximations may be used to increase the efficiency of the forward model.”
BackgroundMethodIntegrating opponent models with Monte-Carlo tree search in poker
Ponsen, M., Gerritsen, G., & Chaslot, G. M. J. B. (2010) · National Conference on Artificial Intelligence · Cited by 40
Knowledge Generation for Improving Simulations in UCT for General Game Playing
Sharma, S., Kobti, Z., & Goodwin, S. D. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-89378-3_5 · Cited by 39 · cited 1×
“The RAVE-max update rule and its stochastic variant δ -RAVE-max were found to improve performance in degenerate cases for the Sum of Switches game (7.3) but were less successful for Go.”
MethodInfluentialEvaluation Function Based Monte-Carlo LOA
Winands, M. H. M., & Björnsson, Y. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_4 · Cited by 39 · cited 4×
“The key difference is that nested Monte Carlo search memorises the best sequence of moves found at each level of the search.”
BackgroundInfluentialLearning is planning: near Bayes-optimal reinforcement learning via Monte-Carlo tree searchPreprint
Asmuth, J., & Littman, M. L. (2012) · arXiv (Cornell University) · doi:10.48550/arxiv.1202.3699 · Cited by 39 · cited 1×
“This is in contrast to plain UCT, which expands all branches indefinitely.”
ResultRevisiting Monte-Carlo Tree Search on a Normal Form Game: NoGo
Chou, C.-H., Teytaud, O., & Yen, S.-J. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-20525-5_8 · Cited by 38 · cited 1×
“This is because some heuristic functions can be slow to compute, so storing the result and limiting the number of nodes that use the heuristic function leads to an increase in the speed of the modified MCTS algorithm.”
BackgroundThe Last-Good-Reply Policy for Monte-Carlo Go
Drake, P. D. (2009) · ICGA Journal · doi:10.3233/icg-2009-32404 · Cited by 37 · cited 10×
“Drake [75] suggests using the Last Good Reply heuristic (6.1.8) to inform simulations, modified by Baier and Drake [17] to include the forgetting of bad moves.”
BackgroundMethodInfluentialScore Bounded Monte-Carlo Tree Search
Cazenave, T., & Saffidine, A. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_9 · Cited by 37 · cited 2×
“Progressive History was shown to perform well for some multi-player board games (7.3), indicating that it may be useful for multi-player games in general.”
BackgroundInfluentialUpper Confidence Trees with Short Term Partial Information
Teytaud, O., & Flory, S. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-20525-5_16 · Cited by 37 · cited 2×
“TDL can learn heuristic value functions to inform the tree policy or the simulation (playout) policy.”
BackgroundEvaluating Root Parallelization in Go
Soejima, Y., Kishimoto, A., & Watanabe, O. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2096427 · Cited by 37 · cited 1×
“Such enhancements might use prior knowledge about a domain or otherwise exploit some unique aspect of it.”
BackgroundOn the Parallelization of Monte-Carlo planning
Gelly, S., Hoock, J., Rimmel, A., Teytaud, O., & Kalemkarian, Y. (2008) · HAL · Cited by 37
Trade-Offs in Sampling-Based Adversarial Planning
Ramanujan, R., & Selman, B. (2011) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v21i1.13472 · Cited by 35 · cited 1×
“See also the use of the history heuristic for improving simulation estimates (6.1.5).”
MethodInfluentialMultiple Tree for Partially Observable Monte-Carlo Tree Search
Auger, D. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-20525-5_6 · Cited by 34 · cited 2×
“Progressive History was shown to perform well for some multi-player board games (7.3), indicating that it may be useful for multi-player games in general.”
BackgroundInfluentialHayward, R. (2012) · ICGA Journal · doi:10.3233/icg-2012-35212 · Cited by 34 · cited 2×
“MoHex has won the 14th and 15th Computer Olympiads to become the reigning Computer Hex world champion [7].”
UCD : Upper confidence bound for rooted directed acyclic graphs
Saffidine, A., Cazenave, T., & Méhat, J. (2011) · Knowledge-Based Systems · doi:10.1016/j.knosys.2011.11.014 · Cited by 33 · cited 2×
“If m = 0 then only actions selected in the tree are used to update nodes, similarly if m is larger than the number of moves in the simulation, this is equivalent to the AMAF algorithm.”
BackgroundMethodMonte Carlo tree search techniques in the game of Kriegspiel
Ciancarini, P., & Favini, G. P. (2009) · Cited by 32
On the huge benefit of decisive moves in Monte-Carlo Tree Search algorithms
Teytaud, F., & Teytaud, O. (2010) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/itw.2010.5593334 · Cited by 32 · cited 3×
“Progressive widening (5.5.1) is employed to limit the degree of exploration by UCB1 to give promising empirical results.”
BackgroundInfluentialMethods of MCTS and the game Arimaa
Kozelek, T. (2009) · Digital Repository (National Repository of Grey Literature) · Cited by 32
Ensemble Monte-Carlo Planning: An Empirical Study
Fern, A., & Lewis, P. (2011) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v21i1.13458 · Cited by 31 · cited 2×
“They consider a class of synthetic spaces in which UCT significantly outperforms minimax.”
Special Issue on Monte Carlo Techniques and Computer Go
Lee, C.-S., Müller, M., & Teytaud, O. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2099154 · Cited by 31
Audouard, P., Chaslot, G. M. J. B., Hoock, J., Pérez, J., Rimmel, A., & Teytaud, O. (2009) · Lecture notes in computer science · doi:10.1007/978-3-642-01129-0_36 · Cited by 29 · cited 2×
“FUSE also uses a game-specific approximation of the reward function, and adjusts the probability of choosing the stopping feature during simulation according to the depth in the tree.”
BackgroundMethodMove-Pruning Techniques for Monte-Carlo Go
Bouzy, B. (2006) · Lecture notes in computer science · doi:10.1007/11922155_8 · Cited by 29 · cited 1×
“At level n > 0 , the simulation uses level n − 1 searches to select each move.”
MethodA UCT agent for Tron: Initial investigations
Samothrakis, S., Rob, D., & Lucas, S. M. (2010) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/itw.2010.5593331 · Cited by 29 · cited 2×
“TDL can learn heuristic value functions to inform the tree policy or the simulation (playout) policy.”
BackgroundSelf-citationHe, S., Wang, Y., Xie, F., Meng, J., Chen, H., Luo, S., Liu, Z., & Zhu, Q. (2008) · Computational Intelligence for Modelling, Control and Automation · doi:10.1109/cimca.2008.82 · Cited by 29 · cited 2×
“This more accurately models the differences in information available to each player than searching a single tree.”
BackgroundParallel Monte-Carlo Tree Search with Simulation Servers
Kato, H., & Takeuchi, I. (2010) · Conference on Technologies and Applications of Artificial Intelligence (TAAI) · doi:10.1109/taai.2010.83 · Cited by 29 · cited 1×
“Second, MHSP’s selection process simply uses average rewards with no exploration term, but initialises the nodes with “optimistic” average values.”
BackgroundCenturio, a General Game Player: Parallel, Java- and ASP-based
Möller, M., Schneider, M., Wegner, M., & Schaub, T. H. (2010) · Künstliche Intell. · doi:10.1007/s13218-010-0077-4 · Cited by 27 · cited 1×
“The RAVE-max update rule and its stochastic variant δ -RAVE-max were found to improve performance in degenerate cases for the Sum of Switches game (7.3) but were less successful for Go.”
MethodInfluentialModelling and evaluation of complex scenarios with the Strategy Game Description Language
Mahlmann, T., Togelius, J., & Yannakakis, G. N. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6032004 · Cited by 27 · cited 1×
“The RAVE-max update rule and its stochastic variant δ -RAVE-max were found to improve performance in degenerate cases for the Sum of Switches game (7.3) but were less successful for Go.”
MethodInfluentialExperiments with Monte Carlo Othello
Hingston, P., & Mašek, M. (2007) · IEEE Congress on Evolutionary Computation (CEC) · doi:10.1109/cec.2007.4425000 · Cited by 26 · cited 2×
“TDL can learn heuristic value functions to inform the tree policy or the simulation (playout) policy.”
BackgroundCADIA-Player : a general game playing agent
Finnsson, H. (2007) · Cited by 25
Monte-Carlo Tree Search in Backgammon
Lishout, F. V., Chaslot, G. M. J. B., & UITERWIJK, J. W. H. M. (2007) · Open Repository and Bibliography (University of Liège) · Cited by 25
Real-Time Solving of Quantified CSPs Based on Monte-Carlo Game Tree Search
Baba, S., Joe, Y., Iwasaki, A., & Yokoo, M. (2012) · Cited by 24
Boosting Active Learning to Optimality: A Tractable Monte-Carlo, Billiard-Based Algorithm
Rolet, P., Sébag, M., & Teytaud, O. (2009) · Lecture notes in computer science · doi:10.1007/978-3-642-04174-7_20 · Cited by 24 · cited 1×
“This section describes some of the characteristics that make MCTS a popular choice of algorithm for a variety of domains, often with notable success.”
BackgroundImproving Monte–Carlo Tree Search in Havannah
Lorentz, R. J. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_10 · Cited by 24 · cited 1×
“Databases of opening move sequences of known utility.”
BackgroundTowards Procedural Strategy Game Generation: Evolving Complementary Unit Types
Mahlmann, T., Togelius, J., & Yannakakis, G. N. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-20525-5_10 · Cited by 24 · cited 1×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundInfluentialMultiple Overlapping Tiles for Contextual Monte Carlo Tree Search
Rimmel, A., & Teytaud, F. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12239-2_21 · Cited by 24 · cited 1×
“Databases of opening move sequences of known utility.”
BackgroundUnderstanding Sampling Style Adversarial Search MethodsPreprint
Ramanujan, R., Sabharwal, A., & Selman, B. (2012) · Conference on Uncertainty in Artificial Intelligence · doi:10.48550/arxiv.1203.4011 · Cited by 24 · cited 1×
“This algorithm is the same as AMAF but also updates nodes that can be reached by permutations of moves in the simulation that preserve the eventual state reached [101].”
MethodCazenave, T. (2007) · Cited by 23
Learning non-random moves for playing Othello: Improving Monte Carlo Tree Search
Robles, D., Rohlfshagen, P., & Lucas, S. M. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6032021 · Cited by 23 · cited 1×
“Another approach called dynamic exploration , proposed by Bourki et al. [25], tunes parameters based on patterns in their Go program M O G O .”
BackgroundSelf-citationNested Monte-Carlo Expression Discovery
Cazenave, T. (2010) · Frontiers in artificial intelligence and applications · doi:10.3233/978-1-60750-606-5-1057 · Cited by 22 · cited 2×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundInfluentialA study on security evaluation methodology for image-based biometrics authentication systems
Yasuhiro, T., Yoshizoe, K., & Imai, H. (2009) · IEEE International Conference on Biometrics: Theory, Applications and Systems (BTAS) · doi:10.1109/btas.2009.5339016 · Cited by 22 · cited 1×
“This idea is similar to First Play Urgency (5.2.1) in that it forces earlier exploitation.”
ResultInfluentialArneson, B., Hayward, R., & Henderson, P. (2009) · ICGA Journal · doi:10.3233/icg-2009-32218 · Cited by 22 · cited 1×
“Databases of opening move sequences of known utility.”
Utilisation de la recherche arborescente Monte-Carlo au Hex
Cazenave, T., & Saffidine, A. (2009) · Revue d intelligence artificielle · doi:10.3166/ria.23.183-202 · Cited by 21
Introduction of a new paraphrase generation tool based on Monte-Carlo sampling
Chevelu, J., Lavergne, T., Lepage, Y., & Moudenc, T. (2009) · Joint Conference of the Annual Meeting of the Association for Computational Linguistics and the International Joint Conference on Natural Language Processing (ACL-IJCNLP) · doi:10.3115/1667583.1667660 · Cited by 21 · cited 2×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundMethodInfluentialMonte-Carlo Tree Search in Crazy Stone
Coulom, R. (2007) · HAL · Cited by 20
Selective search in games of different complexity
Schadd, M. P. D. (2011) · Research Publications (Maastricht University) · doi:10.26481/dis.20110525ms · Cited by 20
Addressing NP-Complete Puzzles with Monte-Carlo Methods 1
Schadd, M. P. D., Winands, M. H. M., van den Herik, H. J., & Aldewereld, H. (2008) · Cited by 20
CadiaPlayer: Search-Control Techniques
Finnsson, H., & Björnsson, Y. (2011) · Künstliche Intell. · doi:10.1007/s13218-010-0080-9 · Cited by 20 · cited 1×
“The RAVE-max update rule and its stochastic variant δ -RAVE-max were found to improve performance in degenerate cases for the Sum of Switches game (7.3) but were less successful for Go.”
MethodInfluentialMONTE-CARLO EXPRESSION DISCOVERY
Cazenave, T. (2012) · International Journal of Artificial Intelligence Tools · doi:10.1142/s0218213012500352 · Cited by 20
Huang, J., Liu, Z., Lu, B., & Xiao, F. (2010) · Conference on Technologies and Applications of Artificial Intelligence (TAAI) · doi:10.1109/taai.2010.38 · Cited by 20 · cited 3×
“Strategy fusion is a problem with determinization techniques, which involves the incorrect assumption that different moves can be chosen from different states in the same information set.”
BackgroundInfluentialMonte-Carlo Proof-Number Search for Computer Go
Saito, J.-T., Chaslot, G. M. J. B., UITERWIJK, J. W. H. M., & van den Herik, H. J. (2007) · Lecture notes in computer science · doi:10.1007/978-3-540-75538-8_5 · Cited by 19 · cited 1×
“Section 4.8.9 discusses this in more detail.”
InfluentialMarcolino, L. S., & Matsubara, H. (2011) · Adaptive Agents and Multi-Agents Systems · doi:10.5555/2030470.2030474 · Cited by 18
A Shogi Program Based on Monte-Carlo Tree Search
Sato, Y., Takahashi, D., & Grimbergen, R. (2010) · ICGA Journal · doi:10.3233/icg-2010-33203 · Cited by 18 · cited 1×
“See also the use of the history heuristic for improving simulation estimates (6.1.5).”
MethodInfluentialBackpropagation Modification in Monte-Carlo Game Tree Search
Xie, F., & Liu, Z. (2009) · Intelligent Information Technology Application · doi:10.1109/iita.2009.331 · Cited by 18 · cited 2×
“Planners for many complex structured domains can be learned with tractable sample complexity if near optimal policies are known.”
BackgroundMethodTime Management for Monte-Carlo Tree Search Applied to the Game of Go
Huang, S.-C., Coulom, R., & Lin, S. (2010) · Conference on Technologies and Applications of Artificial Intelligence (TAAI) · doi:10.1109/taai.2010.78 · Cited by 17 · cited 1×
“In addition, the use of transpositions allows information to be shared between these extra nodes where the state is unchanged.”
BackgroundCazenave, T. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-87608-3_5 · Cited by 17 · cited 2×
“This is because some heuristic functions can be slow to compute, so storing the result and limiting the number of nodes that use the heuristic function leads to an increase in the speed of the modified MCTS algorithm.”
BackgroundMeta Monte-Carlo Tree Search for Automatic Opening Book Generation
Chaslot, G. M. J. B., Hoock, J., Pérez, J., Rimmel, A., Teytaud, O., & Winands, M. H. M. (2009) · Cited by 17
Bandit-Based Genetic Programming
Hoock, J., & Teytaud, O. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12148-7_23 · Cited by 16 · cited 1×
“Both variants improved the playing strength of the program M O G O for 9 × 9 Go when used for generating opening books.”
MethodA Monte-Carlo approach for ghost avoidance in the Ms. Pac-Man game
Tong, B. K.-B., & Sung, C. W. (2010) · International IEEE Consumer Electronics Society's Games Innovations Conference (ICE-GIC) · doi:10.1109/icegic.2010.5716879 · Cited by 16 · cited 1×
“• Combined heuristics can be more powerful than individual heuristics.”
BackgroundMonte-Carlo Tree Search for the game of Scotland Yard
Nijssen, J. A. M., & Winands, M. H. M. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6032002 · Cited by 16 · cited 2×
“MCTS with these enhancements was demonstrated to reduce the number of simulations required to solve seki situations in Go (7.1) and was also shown to be beneficial for the game Connect Four (Section 7.3).”
BackgroundInfluentialGOLOIS WINS PHANTOM GO TOURNAMENT
Cazenave, T., & Borsboom, J. (2007) · ICGA Journal · doi:10.3233/icg-2007-30311 · Cited by 16
Parallel Monte-Carlo Tree Search for HPC Systems
Graf, T., Lorenz, U., Platzner, M., & Schaefers, L. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-23397-5_36 · Cited by 15 · cited 1×
“Their Bayesian MCTS formalism introduces two tree policies: where µ i replaces the average reward of the node with the mean of an extremum (minimax) distribution P i (assuming independent random variables) and: where σ i is the square root of the variance of P i .”
BackgroundMonte-Carlo Style UCT Search for Boolean Satisfiability
Previti, A., Ramanujan, R., Schaerf, M., & Selman, B. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-23954-0_18 · Cited by 15 · cited 2×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundResultInfluentialA Monte-Carlo approach for the endgame of Ms. Pac-Man
Tong, B. K.-B., Man, C., & Sung, C. W. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6031983 · Cited by 14 · cited 1×
“• Combined heuristics can be more powerful than individual heuristics.”
BackgroundEvaluation of Game Tree Search Methods by Game Records
Takeuchi, S., Kaneko, T., & Yamaguchi, K. (2010) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2010.2102022 · Cited by 14 · cited 4×
“Permutation AMAF would also update these nodes.”
BackgroundMethodInfluentialMatsumoto, S., Hirosue, N., Itonaga, K., Yokoo, K., & Futahashi, H. (2010) · Cited by 14
GOLOIS wins Phantom-Go Tournament
Cazenave, T. (2009) · ICGA Journal · doi:10.3233/icg-2009-32110 · Cited by 14
Drake, P. D., & Uurtamo, S. (2007) · Cited by 13
Consistency Modifications for Automatically Tuned Monte-Carlo Tree Search
Berthier, V., Doghmen, H., & Teytaud, O. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-13800-3_9 · Cited by 13 · cited 1×
“See also the bandit-based active learner (4.3.3).”
InfluentialPinson, S., Balbo, F., & Cazenave, T. (2009) · Cited by 13
αβ-based play-outs in Monte-Carlo Tree Search
Winands, M. H. M., & Björnsson, Y. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6031996 · Cited by 13 · cited 1×
“7), which uses a selective twoply α-β search at each playout step [233].”
BackgroundMonte-Carlo Planning for Pathfinding in Real-Time Strategy Games
Naveed, M., Kitchin, D. E., & Crampton, A. (2010) · Huddersfield Research Portal (University of Huddersfield) · Cited by 12
Cazenave, T. (2007) · ICGA Journal · doi:10.3233/icg-2007-30104 · Cited by 12 · cited 1×
“In addition, the use of transpositions allows information to be shared between these extra nodes where the state is unchanged.”
BackgroundMonte-Carlo Opening Books for Amazons
Kloetzer, J. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_12 · Cited by 12 · cited 2×
“Another approach called dynamic exploration , proposed by Bourki et al. [25], tunes parameters based on patterns in their Go program M O G O .”
BackgroundMethodInfluentialOptimal robust expensive optimization is tractable
Rolet, P., Sébag, M., & Teytaud, O. (2009) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/1569901.1570255 · Cited by 12 · cited 1×
“[175], [173], [174] propose the bandit-based active learner (BAAL) method to address the issue of small training sets in applications where data are sparse.”
MethodA comparative study of solvers in Amazons endgames
Kloetzer, J., Iida, H., & Bouzy, B. (2008) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2008.5035665 · Cited by 12 · cited 1×
“Another approach called dynamic exploration , proposed by Bourki et al. [25], tunes parameters based on patterns in their Go program M O G O .”
BackgroundA novel ontology for computer go knowledge management
Lee, C.-S., Wang, M., Hong, T., Chaslot, G. M. J. B., Hoock, J., Rimmel, A., Teytaud, O., & Kuo, Y.-H. (2009) · IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) · doi:10.1109/fuzzy.2009.5277222 · Cited by 12 · cited 1×
“In addition, the use of transpositions allows information to be shared between these extra nodes where the state is unchanged.”
BackgroundApplication of UCT Search to the Connection Games of Hex, Y, *Star, and Renkula!
Raiko, T., & Peltonen, J. (2008) · Cited by 12
Kloetzer, J., Iida, H., & Bouzy, B. (2009) · ICGA Journal · doi:10.3233/icg-2009-32303 · Cited by 11
Cazenave, T. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_5 · Cited by 11 · cited 1×
“RAVE is a softer approach than Cutoff AMAF since exploited areas of the tree will use the accurate statistics more than unexploited areas of the tree.”
MethodAdaptive play in Texas Hold'em Poker
Maîtrepierre, R., Mary, J., & Munos, R. (2008) · Cited by 11
The Dangers of Random Playouts
Browne, C. (2011) · ICGA Journal · doi:10.3233/icg-2011-34105 · Cited by 11 · cited 1×
“We do not guarantee that all cited works have been peer-reviewed or professionally recognised, but have erred on the side of inclusion so that the coverage of material is as comprehensive as possible.”
BackgroundSelf-citationMonte-Carlo Tree Search and Computer Go
Chen, K.-H., Du, D., & Zhang, P. (2009) · Studies in computational intelligence · doi:10.1007/978-3-642-04141-9_10 · Cited by 10 · cited 1×
“In addition, the use of transpositions allows information to be shared between these extra nodes where the state is unchanged.”
BackgroundNested Monte-Carlo Search with AMAF Heuristic
Akiyama, H., Komiya, K., & Kotani, Y. (2010) · Conference on Technologies and Applications of Artificial Intelligence (TAAI) · doi:10.1109/taai.2010.37 · Cited by 10 · cited 1×
“If m = 0 then only actions selected in the tree are used to update nodes, similarly if m is larger than the number of moves in the simulation, this is equivalent to the AMAF algorithm.”
MethodAn Othello evaluation function based on Temporal Difference Learning using probability of winning
Osaki, Y., Shibahara, K., Tajima, Y., & Kotani, Y. (2008) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2008.5035641 · Cited by 10 · cited 2×
“Another approach called dynamic exploration , proposed by Bourki et al. [25], tunes parameters based on patterns in their Go program M O G O .”
BackgroundAn UCT Approach for Anytime Agent-Based Planning
Pellier, D., Bouzy, B., & Métivier, M. (2010) · Advances in intelligent and soft computing · doi:10.1007/978-3-642-12384-9_26 · Cited by 9 · cited 1×
“Relative pruning was found to increase the win rate of the Go program L IN G O against GNU G O 3.8 by approximately 3% [106].”
BackgroundInfluentialUnderstanding sampling-based adversarial search methods
Ramanujan, R., Sabharwal, A., & Selman, B. (2010) · eCommons (Cornell University) · Cited by 9
A new benchmark for artificial intelligence
Kroeker, K. L. (2011) · Communications of the ACM · doi:10.1145/1978542.1978548 · Cited by 9 · cited 1×
“As a domain in which computers are not yet at the level of top human players, Go has become the new benchmark for AI in games [123].”
BackgroundLemmas on partial observation, with application to phantom games
Teytaud, F., & Teytaud, O. (2011) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2011.6032013 · Cited by 9
Combining final score with winning percentage by sigmoid function in Monte-Carlo simulations
Shibahara, K., & Kotani, Y. (2008) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2008.5035638 · Cited by 9 · cited 1×
“See also the use of the history heuristic for improving simulation estimates (6.1.5).”
MethodInfluentialArneson, B., Hayward, R., & Henderson, P. (2010) · ICGA Journal · doi:10.3233/icg-2010-33316 · Cited by 9
Using a monte-carlo approach for bus regulation
Cazenave, T., Balbo, F., & Pinson, S. (2009) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2009.5309838 · Cited by 8
Finding the Needle in the Haystack with Heuristically Guided Swarm Tree Search
Edelkamp, S., Kissmann, P., Sulewski, D., & Messerschmidt, H. (2010) · Cited by 8
A Study of UCT and Its Enhancements in an Artificial Game
Tom, D., & Müller, M. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-12993-3_6 · Cited by 8 · cited 2×
“Asmuth and Littman [9] extend the FSSS technique to Bayesian FSSS (BFS3), which approaches Bayes-optimality as the program’s computational budget is increased.”
BackgroundMethodUCD: Upper Confidence Bound for Rooted Directed Acyclic Graphs
Saffidine, A., Cazenave, T., & Méhat, J. (2010) · Conference on Technologies and Applications of Artificial Intelligence (TAAI) · doi:10.1109/taai.2010.79 · Cited by 8
Learning and knowledge generation in General Games
Sharma, S., Kobti, Z., & Goodwin, S. D. (2008) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2008.5035658 · Cited by 8 · cited 1×
“The RAVE-max update rule and its stochastic variant δ -RAVE-max were found to improve performance in degenerate cases for the Sum of Switches game (7.3) but were less successful for Go.”
MethodInfluentialMulti-Armed Bandit Bayesian Decision Making
McInerney, R. E., & Roberts, S. (2010) · Cited by 8
Repeated-Task Canadian Traveler Problem
Bnaya, Z., Felner, A., Fried, D., Maksin, O., & Shimony, S. E. (2021) · Proceedings of the International Symposium on Combinatorial Search · doi:10.1609/socs.v2i1.18197 · Cited by 7
PARAMETER TUNING BY SIMPLE REGRET ALGORITHMS AND MULTIPLE SIMULTANEOUS HYPOTHESIS TESTING
Bourki, A., Coulm, M., Rolet, P., Teytaud, O., & Vayssière, P. (2010) · International Conference on Informatics in Control, Automation and Robotics (ICINCO) · doi:10.5220/0002949901690173 · Cited by 7 · cited 1×
“However, finding the set of agents with the correct properties (i.e. those that increase playing strength) is computationally intensive.”
BackgroundComputational Experiments with the RAVE Heuristic
Tom, D., & Müller, M. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-17928-0_7 · Cited by 7 · cited 2×
“In order to make the complexity of real-time video games tractable, approximations may be used to increase the efficiency of the forward model.”
BackgroundMethodTo Create Intelligent Adaptive Game Opponent by Using Monte-Carlo for Tree Search
Yang, J., Gao, Y., He, S., Liu, X., Fu, Y., Chen, Y., & Ji, D. (2009) · International Conference on Natural Computation (ICNC) · doi:10.1109/icnc.2009.710 · Cited by 7 · cited 1×
“Wins, draws and losses in simulations are assigned rewards of +1 , 0 and − 1 respectively (as usual), but proven wins and losses are assigned rewards of + ∞ and −∞ .”
BackgroundTree Parallelization of Ary on a Cluster
Méhat, J., & Cazenave, T. (2011) · Cited by 7
Chaslot, G. M. J. B., Hoock, J., Teytaud, F., & Teytaud, O. (2009) · Cited by 6
To Create Adaptive Game Opponent by Using UCT
He, S., Xie, F., Wang, Y., Luo, S., Fu, Y., Yang, J., Liu, Z., & Zhu, Q. (2008) · Computational Intelligence for Modelling, Control and Automation · doi:10.1109/cimca.2008.81 · Cited by 6 · cited 1×
“Wins, draws and losses in simulations are assigned rewards of +1 , 0 and − 1 respectively (as usual), but proven wins and losses are assigned rewards of + ∞ and −∞ .”
BackgroundInvestigating UCT and RAVE: steps towards a more robust method
Tom, D. (2010) · ERA: Education and Research Archive (University of Alberta) · doi:10.7939/r3wx3c · Cited by 6 · cited 2×
“In order to make the complexity of real-time video games tractable, approximations may be used to increase the efficiency of the forward model.”
BackgroundMethodEvaluation of Monte Carlo tree search and the application to Go
Takeuchi, S., Kaneko, T., & Yamaguchi, K. (2008) · IEEE Conference on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2008.5035639 · Cited by 5 · cited 4×
“Permutation AMAF would also update these nodes.”
BackgroundMethodInfluentialA Local Monte Carlo Tree Search Approach in Deterministic Planning
Xie, F., Nakhost, H., & Müller, M. (2011) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v25i1.8039 · Cited by 4 · cited 1×
“This is in contrast to plain UCT, which expands all branches indefinitely.”
Result
Research works printed but not linked: 34
- Entries marked with an asterisk (*) denote support material that does not directly cite MCTS methods
- [6] * --, 'Game Self-Play with Pure Monte-Carlo: The Basin Structure,' Friedrich-Schiller Univ., Jena, Tech. Rep., 2010. -
- [9] J. Asmuth and M. L. Littman, 'Approaching Bayes-optimalilty using Monte-Carlo tree search,' in Proc. 21st Int. Conf. Automat. Plan. Sched. , Freiburg, Germany, 2011. -
- [23] J. Borsboom, J.-T. Saito, G. M. J.-B. Chaslot, and J. W. H. M. Uiterwijk, 'A Comparison of Monte-Carlo Methods for Phantom Go,' in Proc. BeNeLux Conf. Artif. Intell. , Utrecht, Netherlands, 2007, pp. 57-64. -
- [29] --, 'On the Dangers of Random Playouts,' Int. Comp. Games Assoc. J. , vol. 34, no. 1, pp. 25-26, 2010. -
- Brügmann, 1993, Monte Carlo Go
- [37] --, 'Evolving Monte-Carlo Tree Search Algorithms,' Univ. Paris 8, Dept. Inform., Tech. Rep., 2007. -
- [53] G. M. J.-B. Chaslot, L. Chatriot, C. Fiter, S. Gelly, J.-B. Hoock, J. Perez, A. Rimmel, and O. Teytaud, 'Combining expert, offline, transient and online knowledge in Monte-Carlo exploration,' Lab. Rech. Inform. (LRI), Paris, Tech. Rep., 2008. -
- [74] N. G. P. Den Teuling, 'Monte-Carlo Tree Search for the Simultaneous Move Game Tron,' Univ. Maastricht, Netherlands, Tech. Rep., 2011. -
- [77] --, 'Move Ordering vs Heavy Playouts: Where Should Heuristics be Applied in Monte Carlo Go,' in Proc. 3rd North Amer. Game-On Conf. , Gainesville, Florida, 2007, pp. 35-42. -
- [81] M. Enzenberger, M. M¨ uller, B. Arneson, and R. B. Segal, 'Fuego - An Open-Source Framework for Board Games and Go Engine Based on Monte Carlo Tree Search,' IEEE Trans. Comp. Intell. AI Games , vol. 2, no. 4, pp. 259-270, 2010. - · DOI · Paper page
- [85] --, 'Simulation Control in General Game Playing Agents,' in Proc. Int. Joint Conf. Artif. Intell. Workshop Gen. Game Playing , Pasadena, California, 2009, pp. 21-26. -
- [90] S. Gelly, 'A Contribution to Reinforcement Learning; Application to Computer-Go,' Ph.D. dissertation, Univ. Paris-Sud, France, 2007. -
- [107] S.-C. Huang, 'New Heuristics for Monte Carlo Tree Search Applied to the Game of Go,' Ph.D. dissertation, Nat. Taiwan Normal Univ., Taipei, 2011. -
- [120] L. Kocsis, C. Szepesv´ ari, and J. Willemson, 'Improved MonteCarlo Search,' Univ. Tartu, Estonia, Tech. Rep. 1, 2006. -
- [121] S. K¨ onnecke and J. Waldmann, 'Efficient Playouts for the Havannah Abstract Board Game,' Hochschule Technik, Leipzig, Tech. Rep., 2009. -
- [131] * S. Lopez, 'Rybka's Monte Carlo analysis,' 2008. [Online]. Available: http://www.chessbase.com/newsdetail.asp?newsid= 5075 -
- [142] J. M´ ehat and T. Cazenave, 'Ary: A Program for General Game Playing,' Univ. Paris 8, Dept. Inform., Tech. Rep., 2008. - · doi:10.1163/1573-3912_ei2glos_sim_gi_00259
- [143] --, 'Monte-Carlo Tree Search for General Game Playing,' Univ. Paris 8, Dept. Info., Tech. Rep., 2008. -
- [151] K. Q. Nguyen, T. Miyama, A. Yamada, T. Ashida, and R. Thawonmas, 'ICE gUCT,' Int. Comput. Entertain. Lab., Ritsumeikan Univ., Tech. Rep., 2011. -
- [152] J. P. A. M. Nijssen, 'Playing Othello Using Monte Carlo,' Strategies , pp. 1-9, 2007. -
- [154] J. P. A. M. Nijssen and J. W. H. M. Uiterwijk, 'Using Intelligent Search Techniques to Play the Game Khet,' Maastricht Univ., Netherlands, Netherlands, Tech. Rep., 2009. -
- [164] --, 'On the Behavior of UCT in Synthetic Search Spaces,' in Proc. 21st Int. Conf. Automat. Plan. Sched. , Freiburg, Germany, 2011. -
- [176] C. D. Rosin, 'Nested Rollout Policy Adaptation for Monte Carlo Tree Search,' in Proc. 22nd Int. Joint Conf. Artif. Intell. , Barcelona, Spain, 2011, pp. 649-654. -
- [180] A. Saffidine, 'Utilisation dUCT au Hex,' Ecole Normale Super. Lyon, France, Tech. Rep., 2008. -
- [181] --, 'Some Improvements for Monte-Carlo Tree Search, Game Description Language Compilation, Score Bounds and Transpositions,' M.S. thesis, Paris-Dauphine Lamsade, France, 2010. -
- [188] F. C. Schadd, 'Monte-Carlo Search Techniques in the Modern Board Game Thurn and Taxis,' M.S. thesis, Maastricht Univ., Netherlands, 2009. -
- [192] L. Schaefers, M. Platzner, and U. Lorenz, 'UCT-Treesplit - Parallel MCTS on Distributed Memory,' in Proc. 21st Int. Conf. Automat. Plan. Sched. , Freiburg, Germany, 2011. -
- [194] J. Sch¨ afer, 'The UCT Algorithm Applied to Games with Imperfect Information,' Diploma thesis, Otto-Von-Guericke Univ. Magdeburg, Germany, 2008. -
- [196] M. Shafiei, N. R. Sturtevant, and J. Schaeffer, 'Comparing UCT versus CFR in Simultaneous Games,' in Proc. Int. Joint Conf. Artif. Intell. Workshop Gen. Game Playing , Pasadena, California, 2009. -
- [206] J. A. Stankiewicz, 'Knowledge-Based Monte-Carlo Tree Search in Havannah,' M.S. thesis, Maastricht Univ., Netherlands, 2011. -
- [208] N. Sylvester, B. Lohre, S. Dodson, and P. D. Drake, 'A Linear Classifier Outperforms UCT in 9x9 Go,' in Proc. Int. Conf. Artif. Intell. , Las Vegas, Nevada, 2011, pp. 804-808. -
- [231] G. M. J. Williams, 'Determining Game Quality Through UCT Tree Shape Analysis,' M.S. thesis, Imperial Coll., London, 2010. -
- [236] M. H. M. Winands, Y. Bj¨ ornsson, and J.-t. Saito, 'Monte Carlo Tree Search in Lines of Action,' IEEE Trans. Comp. Intell. AI Games , vol. 2, no. 4, pp. 239-250, 2010. - · DOI
Cited by
Mastering the game of Go with deep neural networks and tree search
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016) · Nature · doi:10.1038/nature16961 · Cited by 15,612
“The previous major breakthrough in computer Go, the introduction of MCTS, led to corresponding advances in many other domains; for example, general game-playing, classical planning, partially observed planning, scheduling, and constraint satisfaction 35,36 .”
BackgroundMastering the game of Go without human knowledge
Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., Hubert, T., Baker, L., Lai, M., Bolton, A., et al. (2017) · Nature · doi:10.1038/nature24270 · Cited by 9,084
“Once trained, these networks were combined with a Monte Carlo tree search (MCTS) 13–15 to provide a lookahead search, using the policy network to narrow down the search to highprobability moves, and using the value network (in conjunction with Monte Carlo rollouts using a fast rollout policy) to…”
MethodRecent advances and applications of machine learning in solid-state materials science
Schmidt, J., Marques, M. R. G., Botti, S., & Marques, M. A. L. (2019) · npj Computational Materials · doi:10.1038/s41524-019-0221-0 · Cited by 2,607
SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient
Yu, L., Zhang, W., Wang, J., & Yu, Y. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.10804 · Cited by 2,285
The Arcade Learning Environment: An Evaluation Platform for General Agents
Bellemare, M. G., Naddaf, Y., Veness, J., & Bowling, M. (2013) · Journal of Artificial Intelligence Research · doi:10.1613/jair.3912 · Cited by 2,161
“By expanding the more valuable branches of the tree and carrying out a random simulation at the leaf nodes, UCT is known to perform well in many different settings (Browne et al., 2012).”
Background
Metadata
- Authors
- Browne, C. B., Powley, E., Whitehouse, D., Lucas, S. M., Cowling, P. I., Rohlfshagen, P., Tavener, S., Perez, D., Samothrakis, S., & Colton, S.
- Affiliations
- Imperial College London (GB)
- University of Bradford (GB)
- University of Essex (GB)
- ORCID
- Venue
- IEEE Transactions on Computational Intelligence and AI in Games
- Published
- 2012-03
- Volume
- 4
- Issue
- 1
- Pages
- 1-43
- ISSN
- 1943-068X (ISSN-L)
- 1943-0698
- OpenAlex
- W2126316555
- Semantic Scholar
- c37f1baac3c8ba30250084f067167ac3837cf6fd
- Cited by
- 2,970
- References
- 233
- Citations by year
- 2026: 163 · 2025: 220 · 2024: 244 · 2023: 269 · 2022: 294 · 2021: 308 · 2020: 305 · 2019: 263 · 2018: 230 · 2017: 186 · 2016: 155 · 2015: 117 · 2014: 125 · 2013: 60 · 2012: 28
- FWCI
- 103.94
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- IEEE Transactions on Computational Intelligence and AI in Games (publishedVersion)
- Research Publications (Maastricht University) (publishedVersion)
- Open Access at Essex (University of Essex) (acceptedVersion)
- Research Publications (Maastricht University) (submittedVersion, public-domain)
- Funders
- Engineering and Physical Sciences Research Council (EP/H048588/1, EP/H049061/2, EP/H049061/1, EP/I001964/1)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Social Sciences › Decision Sciences › Management Science and Operations Research › Advanced Bandit Algorithms Research
- Keywords
- Monte Carlo tree search, computer Go, search algorithms, random sampling, game-playing agents, open research questions
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.