The DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index; 5 of the 16 sources its claims cite were checked for support.
A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Open accessAbstract
One program to rule them all: Computers can beat humans at increasingly complex games, including chess and Go. However, these programs are typically constructed for a particular game, exploiting its properties, such as the symmetries of the board on which it is played. Silver et al. developed a program called AlphaZero, which taught itself to play Go, chess, and shogi (a Japanese version of chess) (see the Editorial, and the Perspective by Campbell). AlphaZero managed to beat state-of-the-art programs specializing in these three games. The ability of AlphaZero to adapt to various game rules is a notable step toward achieving a general game-playing system. Science , this issue p. 1140 ; see also pp. 1087 and 1118
Overview
CoChat Summary
TL;DR
AlphaZero is a single reinforcement-learning and MCTS algorithm that, given only game rules, learned superhuman play in chess, shogi, and Go through self-play. It defeated world-champion programs Stockfish, Elmo, and AlphaGo Zero using the same algorithm and network architecture across all three games.
Key findings
- In chess, AlphaZero defeated Stockfish 8, winning 155 games and losing 6 out of 1,000; it also won matches from common human openings and 2016 TCEC opening positions.
- In shogi, AlphaZero defeated Elmo, winning 98.2% of games as black and 91.2% overall; in Go, it defeated AlphaGo Zero, winning 61% of games.
- AlphaZero first outperformed Stockfish after 4 hours (300,000 steps), Elmo after 2 hours (110,000 steps), and AlphaGo Lee after 30 hours (74,000 steps); training ran for 700,000 steps, taking about 9 hours in chess, 12 hours in shogi, and 13 days in Go.
- AlphaZero searched only about 60,000 positions per second in chess and shogi, versus 60 million for Stockfish and 25 million for Elmo, yet still defeated Stockfish with 1/10 thinking time and won 46% against Elmo with 1/100 thinking time.
Methods
- AlphaZero uses a deep neural network \(f_\theta(s)\) that outputs move probabilities \(p\) and a scalar value \(v\), trained by reinforcement learning from self-play starting from random parameters; MCTS uses these outputs to guide search, with 800 simulations per move during training.
- The same algorithm, network architecture, and hyperparameters were used for chess, shogi, and Go, with no domain knowledge except game rules; no opening book, endgame tablebases, or domain-specific heuristics were used.
- Training used 5,000 first-generation TPUs to generate self-play games and 16 second-generation TPUs to train networks; evaluation matches used 3 hours per game plus 15 seconds per move, with opponents on 44 CPU cores and AlphaZero on four first-generation TPUs and 44 CPU cores.
Limitations
- Stated: AlphaZero did not use opening books, endgame tablebases, or domain-specific heuristics; the authors note that some such techniques could further improve performance but were left for future research.
- Stated: hyperparameters were reused across games without game-specific tuning, except for exploration noise and the learning rate schedule; matches against an opening book were low in diversity, and forced randomization increased losses.
- Evident from the text: results cover only three games with separate trained instances, no cross-game transfer is demonstrated, and evaluation compares AlphaZero on TPU hardware against CPU-based opponents, so hardware and time-control differences qualify direct strength comparisons.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 2 Not valid · 8 Not checked · 43 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Apart from the representation of positions and actions described above, AlphaZero uses the same network architecture as AlphaGo Zero ( 9 ), briefly recapitulated here.”
What the sources say
- Supportedcites [9] source
The Neural Network Architecture Comparison (p. 29) identifies AlphaGo Zero’s architecture as a residual tower with policy and value heads. The citation supplies the AlphaGo Zero architecture that the citing authors say AlphaZero uses; the comparison to AlphaZero is their own statement. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“and in Go, AlphaZero first outperformed AlphaGo Lee ( 9 ) after 30 hours (74,000 steps).”
What the sources say
- Partly supportedcites [9] sourceExperimental — accuracy not yet measured
Section 2, p. 7 reports that AlphaGo Zero outperformed AlphaGo Lee, but says this happened after 36 hours, not 30. The passage does not give 74,000 steps, and it names the system AlphaGo Zero rather than AlphaZero. (LLM, conf=0.90)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The training algorithm achieved similar performance in all independent runs (see fig. S3), suggesting that the high performance of AlphaZero's training algorithm is repeatable.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“AlphaZero also defeated Stockfish when given 1 / 10 as much thinking time as its opponent (i.e. searching ~ 1 / 10 , 000 as many positions),”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“and won 46% of games against Elmo when given 1 / 100 as much time (i.e. searching ~ 1 / 40 , 000 as many positions), see Fig. 2.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
23 more claims
Create a free account to see all 56 claims and to verify the 3 references still open.
- Probably not a substantial claim. 10
- It may state two separate claims as one. 23
- The quote only partly states this claim. 1
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 source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Games have historically served as highly effective metrics for evaluating progress in deep reinforcement learning research”
From BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games (2024) · cites this paper as Silver et al., 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In 2017, AlphaGo, using the same learning algorithms in TD-Gammon and a billion times more computing power, soundly beat Ke Jie, the world Go champion, in a game that is to chess what chess is to backgammon in complexity (Silver et al., 2018).”
From Large Language Models and the Reverse Turing Test (2023) · cites this paper as Silver et al., 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Most of the attention was on who won, but more important was how it won: AlphaGo was creative and had discovered new moves and board positions that were outside human experience.”
From Large Language Models and the Reverse Turing Test (2023) · cites this paper as Silver et al., 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The primary method for applying more computation at inference, common in the AlphaZero family of agents2,72, is scaling the tree search and running multiple independent search attempts.”
From Olympiad-level formal mathematical reasoning with reinforcement learning (2026) · cites this paper as [2]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Modern AI can write computer programs (Li et al., 2022) or play chess at super-human level (Silver et al., 2018), but the ability of AI to perceive and act in the world remains far below human level.”
From Scaling Instructable Agents Across Many Simulated Worlds (2024) · cites this paper as Silver et al., 2018
1 more claim about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
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
Self-citationAdvances in neural information processing systems 7
(1996) · Computers & Mathematics with Applications · doi:10.1016/s0898-1221(96)90463-0 · Cited by 14,281
Identity mappings in deep residual networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_38 · Cited by 10,741
Identity Mappings in Deep Residual Networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · arXiv · doi:10.48550/arxiv.1603.05027 · Cited by 10,665
Mastering 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
Self-citationIn-Datacenter Performance Analysis of a Tensor Processing Unit
Jouppi, N. P., Young, C., Patil, N., Patterson, D., Agrawal, G., Bajwa, R., Bates, S., Bhatia, S., Boden, N., Borchers, A., et al. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080246 · Cited by 4,537
Encyclopedia of Artificial Intelligence
(2009) · IGI Global eBooks · doi:10.4018/978-1-59904-849-9 · Cited by 1,396
Campbell, M., Hoane, A. J., & Hsu, F.-H. (2002) · Artificial Intelligence · doi:10.1016/s0004-3702(01)00129-1 · Cited by 1,226
Lim, T.-S., Loh, W.-Y., & Shih, Y.-S. (2000) · Machine Learning · doi:10.1023/a:1007608224229 · Cited by 1,133
An analysis of alpha-beta pruning
Knuth, D. E., & Moore, R. W. (1975) · Artificial Intelligence · doi:10.1016/0004-3702(75)90019-3 · Cited by 1,004
Research works printed but not linked: 7
- C. J. Maddison A. Huang I. Sutskever D. Silver paper presented at the International Conference on Learning Representations 2015 San Diego CA 7 to 9 May 2015.
- D. N. L. Levy M. Newborn How Computers Play Chess (Ishi Press 2009).
- W. Steinitz The Modern Chess Instructor (Edition Olms 1990).
- O. Arenz “Monte Carlo chess ” master’s thesis Technische Universität Darmstadt (2012).
- 10.1023/a:1007634325138 · DOI
- T. Kaneko K. Hoki in Advances in Computer Games – 13th International Conference ACG 2011 Revised Selected Papers Tilburg Netherlands 20 to 22 November 2011 (Springer 2012) pp. 158–169.
- R. Ramanujan A. Sabharwal B. Selman in Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence (UAI 2010) Catalina Island CA 8 to 11 July (AUAI Press 2010).
Cited by
Items 1,101–1,200 of 3,622
Wang, Q., Hao, Y., & Zhang, J. (2023) · Journal of King Saud University - Computer and Information Sciences · doi:10.1016/j.jksuci.2023.101787 · Cited by 17
Deep reinforcement learning for continuous wood drying production line control
Tremblay, F.-A., Durand, A., Morin, M., Marier, P., & Gaudreault, J. (2023) · Computers in Industry · doi:10.1016/j.compind.2023.104036 · Cited by 14
Automated algorithm selection using meta-learning and pre-trained deep convolution neural networks
Dagan, I., Vainshtein, R., Katz, G., & Rokach, L. (2023) · Information Fusion · doi:10.1016/j.inffus.2023.102210 · Cited by 15
The Social Machine: Artificial Intelligence (AI) Approaches to Theory of Mind
Nebreda, A., Shpakivska‐Bilán, D., Cámara, C., & Susi, G. (2023) · Logic, argumentation & reasoning · doi:10.1007/978-3-031-46742-4_22 · Cited by 19
Gürses, Y., Buyukdemirci, K., & Yıldız, Y. (2024) · IEEE Control Systems Letters · doi:10.1109/lcsys.2024.3349511 · Cited by 18
“Although rare, they are valuable for the learning process, but they lead to environments with sparse rewards.”
BackgroundWu, G., Zhang, D., Miao, Z., Bao, W., & Cao, J. (2024) · Electronics · doi:10.3390/electronics13071281 · Cited by 16
Tuzsus, D., Brands, A. M. A., Pappas, I., & Peters, J. (2024) · Computational Brain & Behavior · doi:10.1007/s42113-024-00202-y · Cited by 12
“While deep RL agents show superhuman ability in games like Go, Shogi, Chess and Atari games (Mnih et al., 2015; Silver et al., 2017, 2018) they fail to perform better than chance level on a standard T-maze task from animal learning (Wauthier et al., 2021).”
Improving algorithmic trading consistency via human alignment and imitation learning
Huang, Y., Zhou, C., Cui, K., & Lu, X. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.124350 · Cited by 14
Reinforcement learning in cold atom experiments
Reinschmidt, M., Fortágh, J., Günther, A., & Volchkov, V. (2024) · Nature Communications · doi:10.1038/s41467-024-52775-8 · Cited by 33
Göttl, Q., Pirnay, J., Burger, J., & Grimm, D. G. (2024) · Computers & Chemical Engineering · doi:10.1016/j.compchemeng.2024.108975 · Cited by 23
Optimizing the hyper-parameters of deep reinforcement learning for building control
Li, S., Su, S., & Lin, X. (2025) · Building Simulation · doi:10.1007/s12273-025-1233-y · Cited by 13
“It has achieved significant success in fields such as gaming (e.g., AlphaGo) (Silver et al. 2018), autonomous driving (Wu et al. 2023), and robotics (Levine et al. 2018).”
Deep learning-based image reconstruction for electrical capacitance tomography
Peng, L., Yang, Y., Li, Y., Zhang, M., Wang, H., & Yang, W. (2025) · Measurement Science and Technology · doi:10.1088/1361-6501/add8ad · Cited by 16
“The success of AlphaGo (Silver et al 2016) and AlphaZero (Silver et al 2018) may provide some new ideas for ECT image reconstruction based on deep learning, especially deep reinforcement learning.”
AI in Neurology: Everything, Everywhere, All at Once Part 1: Principles and Practice
Rizzo, M., & Dawson, J. D. (2025) · Annals of Neurology · doi:10.1002/ana.27225 · Cited by 20
No-boarding buses: Synchronisation for efficiency
Saw, V.-L., & Chew, L. Y. (2020) · PLoS ONE · doi:10.1371/journal.pone.0230377 · Cited by 14
“The idea there is similar to the sensational AlphaZero programme which successfully dominated three different games of Go, Chess, and Shogi by reinforcement learning [68].”
MethodReview of the current state of digital image analysis in breast pathology
Chang, M. C., & Mrkonjic, M. (2020) · The Breast Journal · doi:10.1111/tbj.13858 · Cited by 12
A Performance-Based Start State Curriculum Framework for Reinforcement Learning
Wöhlke, J., Schmitt, F., & van Hoof, H. (2020) · Adaptive Agents and Multi-Agents Systems · doi:10.65109/llzi8531 · Cited by 23
“Although reinforcement learning (RL) has shown remarkable success in playing games with super-human performance [13, 22, 36] as well as mastering robotic locomotion [35] or manipulation tasks [1, 21], sparse reward problems still present a challenge.”
BackgroundMaffettone, P. M., Lynch, J. K., Caswell, T. A., et al. (2021) · Machine Learning Science and Technology · doi:10.1088/2632-2153/abc9fc · Cited by 12
Policy Evaluation and Seeking for Multiagent Reinforcement Learning via Best Response
Yan, R., Duan, X., Shi, Z., Zhong, Y., Marden, J. R., & Bullo, F. (2021) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2021.3085171 · Cited by 13
Learning manipulation skills with demonstrations for the swing process control of dredgers
Wei, C., Bai, H., Wei, Y., Ji, Z., & Liu, Z. (2022) · Ocean Engineering · doi:10.1016/j.oceaneng.2022.110545 · Cited by 15
The Science of the Deal: Optimal Bargaining on eBay Using Deep Reinforcement Learning
Green, E. A., & Plunkett, E. B. (2022) · ACM Conference on Economics and Computation (EC) · doi:10.1145/3490486.3538373 · Cited by 11
“A striking feature of recent game-playing RL agents, such as those for chess and Go [18], is that they learn to beat the best humans without ever playing a human.”
BackgroundHow to incorporate biological insights into network models and why it matters
Timón, L. B., Ekelmans, P., Kraynyukova, N., Rose, T., Busse, L., & Tchumatchenko, T. (2022) · The Journal of Physiology · doi:10.1113/jp282755 · Cited by 18
“Today, some of the trained ANNs that outperform humans on specific tasks (Silver et al. 2018; 2017) do not consider any detailed ion channel dynamics underlying a neural spike, nor do they learn according to the biologically reported local laws of synaptic plasticity.”
BackgroundGuo, M., Mukherjee, M., Lloret, J., Li, L., Guan, Q., & Ji, F. (2022) · Digital Communications and Networks · doi:10.1016/j.dcan.2022.09.020 · Cited by 13
Intelligent Control of Groundwater in Slopes with Deep Reinforcement Learning
Biniyaz, A., Azmoon, B., & Liu, Z. “. (2022) · Sensors · doi:10.3390/s22218503 · Cited by 10
Deep learning and computer vision techniques for microcirculation analysis: A review
Helmy, M., Truong, T. T., Jul, E., & Ferreira, P. A. V. (2022) · Patterns · doi:10.1016/j.patter.2022.100641 · Cited by 18
“Reinforcement learning is common in game development for games like chess and Go [42, 43].”
BackgroundAnalyzing neural network behavior through deep statistical model checking
Gros, T. P., Hermanns, H., Hoffmann, J., Klauck, M., & Steinmetz, M. (2022) · International Journal on Software Tools for Technology Transfer · doi:10.1007/s10009-022-00685-9 · Cited by 12
“Neural networks (NN) have recently been applied with dramatic successes to the learning of action policies in large transition systems, fromAtari games [59] over Go and Chess [67] to Rubik’s cube [1].”
MethodQuantum circuit compilation for nearest-neighbor architecture based on reinforcement learning
Li, Y., Liu, W., Li, M., & Li, Y. (2023) · Quantum Information Processing · doi:10.1007/s11128-023-04050-w · Cited by 18
Risk-aware shielding of Partially Observable Monte Carlo Planning policies
Mazzi, G., Castellini, A., & Farinelli, A. (2023) · Artificial Intelligence · doi:10.1016/j.artint.2023.103987 · Cited by 17
Liu, L., & Yang, J. H. (2023) · Computers & Industrial Engineering · doi:10.1016/j.cie.2023.109564 · Cited by 14
End-to-end path planning for homogeneous temperature fields in additive manufacturing
Sideris, I., Duncan, S., Fabbri, M., Crivelli, F., Afrasiabi, M., & Bambach, M. (2024) · Journal of Materials Processing Technology · doi:10.1016/j.jmatprotec.2024.118364 · Cited by 14
“It is particularly well-suited for problems with large state and action spaces, such as the game of Go (Silver et al., 2016) and chess (Silver et al., 2018), due to its stochastic, online, anytime nature.”
Autonome Entdeckung in den chemischen Wissenschaften, Teil I: Fortschritt
Coley, C. W., Eyke, N. S., & Jensen, K. F. (2019) · Angewandte Chemie · doi:10.1002/ange.201909987 · Cited by 12
Biologically Inspired Visual System Architecture for Object Recognition in Autonomous Systems
Malowany, D., & Guterman, H. (2020) · Algorithms · doi:10.3390/a13070167 · Cited by 15
“In addition, recent successes of deep reinforcement learning (RL) techniques have raised widespread interest in their potential for solving practical problems, with most famous cases being in the field of autonomous driving [29] and game playing [30,31].”
BackgroundThe Go Transformer: Natural Language Modeling for Game Play
Ciolino, M., Kalin, J. D., & Noever, D. A. (2020) · International Conference on Artificial Intelligence for Industries (AI4I) · doi:10.1109/ai4i49448.2020.00012 · Cited by 14
“Compared to MCTS itself, the output does not obviously rely on look-ahead search or playing out forward scenarios to 50-60 moves to evaluate rewards.”
MethodInfluentialTracking the Race Between Deep Reinforcement Learning and Imitation Learning
Gros, T. P., Höller, D., Hoffmann, J., & Wolf, V. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-59854-9_2 · Cited by 11
Anchor: The achieved goal to replace the subgoal for hierarchical reinforcement learning
Li, R., Cai, Z., Huang, T., & Zhu, W. (2021) · Knowledge-Based Systems · doi:10.1016/j.knosys.2021.107128 · Cited by 12
Gao, Y., & Wu, L. (2021) · Electronics · doi:10.3390/electronics10131533 · Cited by 13
“The AlphaGo family, including AlphaGo, AlphaGo Zero, and AlphaZero [1–3], have had great success in many complex board games such as Go and chess.”
Towards an AI-powered Player in Cyber Defence Exercises
Meier, R., Lavrenovs, A., Heinaaro, K., Gambazzi, L., & Lenders, V. (2021) · doi:10.23919/cycon51939.2021.9467801 · Cited by 14
Multiplayer AlphaZeroPreprint
Petosa, N., & Balch, T. R. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.13012 · Cited by 12
“Our biggest limitation was compute - DeepMind trained AlphaZero to master Go in 13 days with 5000 first generation TPUs and 16 second-generation TPUs, but with our hardware, that result would take years to replicate [8].”
MethodRobust-Adaptive Control of Linear Systems: beyond Quadratic CostsPreprint
Leurent, E., Efimov, D. V., & Maillard, O.-A. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.10816 · Cited by 12
“Despite the recent successes of Reinforcement Learning in cracking many hard problems (e.g. Mnih et al., 2015; Silver et al., 2018), this tool has hardly been applied in real industrial issues.”
Simulation-Driven Balancing of Competitive Game Levels With Reinforcement Learning
Rupp, F., Eberhardinger, M., & Eckert, K. (2024) · IEEE Transactions on Games · doi:10.1109/tg.2024.3399536 · Cited by 11
“In the games community, RL has been widely applied to play games [29], [30], but recently, it has also been applied to PCG [12], [31], [32], [33].”
The future of fundamental science led by generative closed-loop artificial intelligence
Zenil, H., Tegner, J. N., Abrahão, F. S., Lavin, A., Kumar, V., Frey, J. G., Weller, A., Soldatova, L., Bundy, A., Jennings, N. R., et al. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2026.1678539 · Cited by 22
“So far, AI’s most prominent successes have occurred in well-bounded domains with clear metrics of performance (McCarthy, 2007), such as games and bioinformatics (Silver et al., 2018; Stokes et al., 2020; Alipanahi et al., 2015; Senior et al., 2020).”
Neuromorphic Engineering Needs Closed-Loop Benchmarks
Milde, M. B., Afshar, S., Xu, Y., Marcireau, A., Joubert, D., Ramesh, B., Bethi, Y., Ralph, N. O., Arja, S., Dennler, N., et al. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2022.813555 · Cited by 11
“The Reinforcement Learning (RL) community relies on (simple) tasks that encompass both perception and action, such as Chess (Silver et al., 2018), Atari games (Mnih et al., 2013), or Go (Silver et al., 2017).”
BackgroundMethodEfficient graph neural architecture search using Monte Carlo Tree search and prediction network
Deng, T., & Wu, J. (2022) · Expert Systems with Applications · doi:10.1016/j.eswa.2022.118916 · Cited by 12
Socialni odnosi, animizem, antropomorfizem in interakcija z UI
Babić, N. Č. (2022) · doi:10.18690/um.ff.11.2022.12 · Cited by 11
Reinforcement Learning-Based Inertia and Droop Control for Wind Farm Frequency Regulation
Liang, Y., Sun, L. Z., & Zhao, X. (2022) · IEEE Power Engineering Society General Meeting · doi:10.1109/pesgm48719.2022.9917075 · Cited by 11
“forcement learning (DRL) in modeling computational challenging decision-making problems such as Atari [11], Go [12], and StarCraft [13].”
MethodUsing Simulation Optimization to Improve Zero-shot Policy Transfer of Quadrotors
Gronauer, S., Kissel, M., Sacchetto, L., Korte, M., & Diepold, K. J. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981229 · Cited by 15
PBQ-Enhanced QUIC: QUIC with Deep Reinforcement Learning Congestion Control Mechanism
Zhang, Z., Li, S., Ge, Y., Xiong, G., Zhang, Y., & Xiong, K. (2023) · Entropy · doi:10.3390/e25020294 · Cited by 14
PiCor: Multi-Task Deep Reinforcement Learning with Policy Correction
Bai, F., Zhang, H., Tao, T., Wu, Z., Wang, Y., & Xu, B. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i6.25825 · Cited by 34
“…neural networks (LeCun, Bengio, and Hinton 2015) shows powerful capabilities on various single-task decision-making problems, including board games (Silver et al. 2017, 2018), video games (Mnih et al. 2015; Vinyals et al. 2019), robotic control (Levine et al. 2016; Rajeswaran et al. 2018;…”
BackgroundPPLC: Data-driven offline learning approach for excavating control of cutter suction dredgers
Wei, C., Wang, H., Bai, H., et al. (2023) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2023.106708 · Cited by 11
Simulation Optimization in the New Era of AI
Peng, Y., Chen, C., & Fu, M. C. (2023) · doi:10.1287/educ.2023.0264 · Cited by 12
The Probabilistic Liveness Decision Method of Unbounded Petri Nets Based on Machine Learning
Qi, H., Wang, J., Yan, C., & Jiang, C. (2023) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2023.3323342 · Cited by 12
“Breakthroughs have been obtained in many areas, such as AlphaGo [35], image recognition [36], fraud detection [37], and so on.”
BackgroundEfficient DRL-Based Congestion Control With Ultra-Low Overhead
Han, T., Liao, X., Zeng, C., Sun, D., Zhang, J., & Chen, K. (2023) · IEEE/ACM Transactions on Networking · doi:10.1109/tnet.2023.3330737 · Cited by 11
“Armed with deep neural networks, deep reinforcement learning plays the core role in achieving super-human performance in many games and real-world decision-making tasks [1], [3].”
BackgroundChen, Q., Du, Y., Jiao, Y., Lu, X., Wu, X., & Zhao, Q. (2024) · Quantum Science and Technology · doi:10.1088/2058-9565/ad420a · Cited by 24
Assessing the risk of takeover catastrophe from large language models
Baum, S. D. (2024) · Risk Analysis · doi:10.1111/risa.14353 · Cited by 11
“However, currently, the cutting edge of LLMs is driven not by intelligence amplification but instead by major industrial R&D, including massive investments in computing power and advancements in both hardware and software (Scharre, 2024).”
Economical Electric Vehicle Charging Scheduling via Deep Imitation Learning
Huang, Z.-H., Wang, J., Fan, X., Yue, R., Xiang, C., & Gao, S. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3434734 · Cited by 13
“Reinforcement learning (RL), particularly deep RL (DRL), which integrates RL with deep learning, is extensively utilized for decision-making in nonlinear and stochastic systems and has shown significant success in complex scenarios [20].”
BackgroundSolving the Vehicle Routing Problem with Stochastic Travel Cost Using Deep Reinforcement Learning
Cai, H., Xu, P., Tang, X., & Gan, L. (2024) · Electronics · doi:10.3390/electronics13163242 · Cited by 17
State-of-the-Art of Deep Learning in Multidisciplinary Optical Coherence Tomography Applications
Kalupahana, T. D., Kahatapitiya, N. S., Kamalathasan, D., Wijesinghe, R. E., Silva, B. N., & Wijenayake, U. (2024) · IEEE Access · doi:10.1109/access.2024.3492389 · Cited by 11
Enhancing Adaptability and Efficiency of Task Offloading by Broad Learning in Industrial IoT
Chi, J., Zhou, X., Xiao, F., Qiu, T., & Chen, C. L. P. (2024) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2024.3493053 · Cited by 11
Predicting and publishing accurate imbalance prices using Monte Carlo Tree Search
Pavirani, F., Gompel, J. V., Madahi, S. S. K., Claessens, B., & Develder, C. (2025) · Applied Energy · doi:10.1016/j.apenergy.2025.125944 · Cited by 11
Gao, Q., Yang, H., Theisen, M. F., & Schweidtmann, A. M. (2025) · Computers & Chemical Engineering · doi:10.1016/j.compchemeng.2025.109192 · Cited by 12
Lu, Y. (2025) · Scientific Reports · doi:10.1038/s41598-025-10512-1 · Cited by 30
Multi-objective deep learning: Taxonomy and survey of the state of the art
Peitz, S., & Hotegni, S. S. (2025) · Machine Learning with Applications · doi:10.1016/j.mlwa.2025.100700 · Cited by 28
Biasing MCTS with Features for General Games
Soemers, D. J. N. J., Piette, É., & Browne, C. (2019) · IEEE Congress on Evolutionary Computation (CEC) · doi:10.1109/cec.2019.8790141 · Cited by 13
“This can easily lead to excessively large numbers of outputs in some games, such as 11, 259 outputs in Shogi [3].”
BackgroundInfluentialHybrid Online and Offline Reinforcement Learning for Tibetan Jiu Chess
Li, X., Lv, Z., Wu, L., Zhao, Y., & Xu, X. (2020) · Complexity · doi:10.1155/2020/4708075 · Cited by 11
“…(UCT) algorithm [17–20], temporal difference algorithm [21], the deep learning model combined with UCT search algorithm [22], and DQN al-gorithm [23, 24], and they have also achieved quite good results in computer Go game, which shows that the idea of deep reinforcement learning algorithm can adapt…”
MethodDecentralized TD Tracking with Linear Function Approximation and its Finite-Time Analysis
Wang, G., Lu, S., Giannakis, G. B., et al. (2020) · neural information processing systems · Cited by 16
de Orio, R. L., Ender, J., Fiorentini, S., et al. (2021) · Micromachines · doi:10.3390/mi12040443 · Cited by 11
“Big RL breakthroughs were achieved lately using games like chess or Go [32], but this type of learning algorithm has also been successfully applied for physics-based problems.”
An Interactive Robot Platform for Introducing Reinforcement Learning to K-12 Students
Zhang, Z., Willner-Giwerc, S., Sinapov, J., et al. (2021) · Advances in intelligent systems and computing · doi:10.1007/978-3-030-82544-7_27 · Cited by 11
An Adaptive Threshold for the Canny Algorithm With Deep Reinforcement Learning
Choi, K.-H., & Ha, J.-E. (2021) · IEEE Access · doi:10.1109/access.2021.3130132 · Cited by 12
Approximate Exploitability: Learning a Best Response
Timbers, F., Bard, N., Lockhart, E., Lanctot, M., Schmid, M. S., BURCH, N. R., Schrittwieser, J., Hubert, T., & Bowling, M. (2022) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2022/484 · Cited by 37
Self-citationAn, G. C. (2022) · Frontiers in Systems Biology · doi:10.3389/fsysb.2022.901159 · Cited by 11
“The solution to this problem can be seen in perhaps the most dramatic successes of modern AI: The game-playing AIs from Deep Mind which have mastered Go and other games (Silver et al., 2017; Silver et al., 2018; Vinyals et al., 2019).”
BackgroundExplainable artificial intelligence for assault sentence prediction in New Zealand
Rodger, H., Lensen, A., & Betkier, M. (2022) · Journal of the Royal Society of New Zealand · doi:10.1080/03036758.2022.2114506 · Cited by 11
Shin, D. J., Choi, G., Hong, C., & Han, J. W. (2023) · Molecular Catalysis · doi:10.1016/j.mcat.2023.113096 · Cited by 13
Makropoulos, C., & Bouziotas, D. (2023) · Journal of Hydroinformatics · doi:10.2166/hydro.2023.168 · Cited by 11
Growing Robot Navigation Based on Deep Reinforcement Learning
Ataka, A., & Sandiwan, A. P. (2023) · International Conference on Control, Automation and Robotics (ICCAR) · doi:10.1109/iccar57134.2023.10151740 · Cited by 12
“On another note, a reinforcement learning has shown to work successfully in solving various challenging problems, ranging from games such as Atari [14], [15], to chess, Go, and Dota2 [16]-[17].”
BackgroundRouting optimization with Monte Carlo Tree Search-based multi-agent reinforcement learning
Wang, Q., & Hao, Y. (2023) · Applied Intelligence · doi:10.1007/s10489-023-04881-1 · Cited by 12
“Deepmind has revolutionized artificial intelligence by showing that deep learning and reinforcement learning [7] can solve some CO problems since playing Go [8], resource allocation [9], and matrix decomposition [10] are about searching for solutions across large combinatorial spaces.”
BackgroundWill recent advances in AI result in a paradigm shift in Astrobiology and SETI?
Gale, J., Wandel, A., & Hill, H. (2019) · International Journal of Astrobiology · doi:10.1017/s1473550419000260 · Cited by 17
Shtanko, A., & Kulik, S. D. (2019) · Journal of Physics Conference Series · doi:10.1088/1742-6596/1406/1/012014 · Cited by 11
“As such the breakthrough algorithm capable of playing go, for example, was later expanded to be able to play chess and shogi [22].”
BackgroundMobile Networks for Computer Go
Cazenave, T. (2020) · IEEE Transactions on Games · doi:10.1109/tg.2020.3041375 · Cited by 12
“Later Alpha Zero successfully applied the same principles to the games of Chess and Shogi [5].”
BackgroundMethodHindsight and Sequential Rationality of Correlated Play
Morrill, D., D'Orazio, R., Sarfati, R., Lanctot, M., Wright, J. R., Greenwald, A., & Bowling, M. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i6.16702 · Cited by 14
“Algorithms to approximate maximin or Nash equilibrium strategies have fueled major successes in the autonomous play of human-scale games like Go (Silver et al. 2016, 2017b, 2018), chess and shogi (Silver et al. 2017a, 2018), poker (Moravčı́k et al. 2017; Brown and Sandholm 2018, 2019) and…”
BackgroundSelf-citationMolecular dynamics simulation of nanofilament breakage in neuromorphic nanoparticle networks
Wu, W., Pavloudis, T., Verkhovtsev, A. V., Solov’yov, A. V., & Palmer, R. E. (2022) · Nanotechnology · doi:10.1088/1361-6528/ac5e6d · Cited by 11
From Analog to Digital Computing: Is Homo sapiens’ Brain on Its Way to Become a Turing Machine?
Danchin, A., & Fenton, A. A. (2022) · Frontiers in Ecology and Evolution · doi:10.3389/fevo.2022.796413 · Cited by 12
“They continue to be developed by improving the structures and functioning of various networks (Silver et al., 2018; Czech et al., 2020).”
BackgroundMastering construction heuristics with self-play deep reinforcement learning
Wang, Q., He, Y., & Tang, C. (2022) · Neural Computing and Applications · doi:10.1007/s00521-022-07989-6 · Cited by 11
“AlphaGo series (including deep learning, reinforcement learning, and Monte Carlo tree search [13]) [14–18], for example, have achieved state-of-the-art performance in the Go game.”
Risk of Stochastic Systems for Temporal Logic Specifications
Lindemann, L., Jiang, L., Matni, N., & Pappas, G. J. (2023) · ACM Transactions on Embedded Computing Systems · doi:10.1145/3580490 · Cited by 17
“, robotic manipulation [44], solving games [55, 74], and autonomous driving [19].”
BackgroundDeep Reasoning Networks: Thinking Fast and SlowPreprint
Chen, D., Bai, Y., Zhao, W., Ament, S., Gregoire, J. M., & Gomes, C. P. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.00855 · Cited by 29
“…reasoning ( System 2 ), such as playing Go or crystal structure phase mapping, pure machine learning approaches have to be complemented with reasoning algorithms, such as Monte Carlo tree search [Anthony et al., 2017, Silver et al., 2016, 2018], or mixed-integer programming [Ermon et al., 2015].”
Deep reinforcement learning approach to MIMO precoding problem: Optimality and RobustnessPreprint
Lee, H., Girnyk, M. A., & Jeong, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.16646 · Cited by 19
“AlphaZero [11] is an example of this category and uses a Monte-Carlo tree search model to predict the next series of state transitions, which in turn provides the value of each action before its execution.”
MethodImbrie, A., Kania, E. B., & Laskai, L. (2020) · doi:10.51593/20190047 · Cited by 21
López‐Pérez, D., Domenico, A. D., Piovesan, N., Baohongqiang, H., Geng, X., Qitao, S., & Debbah, M. A. (2021) · arXiv (Cornell University) · Cited by 11
INT: An Inequality Benchmark for Evaluating Generalization in Theorem ProvingPreprint
Wu, Y., Jiang, A. Q., Ba, J. L., & Grosse, R. (2021) · arXiv (Cornell University) · Cited by 11
Künstliche Intelligenz – multiskale und cross‐domäne Synergien von Raumfahrt und Bauwesen
Kraus, M. A., Drass, M., Hörsch, B., Schneider, J., & Kaufmann, W. (2022) · Wiley eBooks · doi:10.1002/9783433610879.ch9 · Cited by 12
Ab-initio quantum chemistry with neural-network wavefunctionsPreprint
Hermann, J., Spencer, J. S., Choo, K., et al. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2208.12590 · Cited by 11
“This is known as ‘self-play’ and has been the basis for many advances in symmetric games such as chess 32,33 .”
BackgroundBalancing of competitive two-player Game Levels with Reinforcement Learning
Rupp, F., Eberhardinger, M., & Eckert, K. (2023) · 2023 IEEE Conference on Games (CoG) · doi:10.1109/cog57401.2023.10333248 · Cited by 10
“In the games community, RL has been widely applied to play games [20], [21], but recently, it has also been used for PCG [9], [22], [23].”
MethodQi, J., Hu, M., Xing, L., & Su, Z. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.128502 · Cited by 11
Optimal Control in Soft and Active Matter
Alvarado, J., Teich, E. G., Sivak, D. A., & Bechhoefer, J. (2026) · Annual Review of Condensed Matter Physics · doi:10.1146/annurev-conmatphys-031324-031350 · Cited by 42
Hügle, T. (2022) · Annals of the Rheumatic Diseases · doi:10.1136/annrheumdis-2022-222141 · Cited by 10
Sample Efficient Reinforcement Learning Using Graph-Based Memory Reconstruction
Kang, Y., Zhao, E., Zang, Y., Li, L., Li, K., Tao, P., & Xing, J. (2023) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2023.3268612 · Cited by 16
“R EINFORCEMENT learning (RL) agents have achieved superhuman performances in various game environments, such as Atari 2600 games [1], [2], Go [3], [4], [5], and Star-Craft [6].”
Reinforcement learning based agents for improving layouts of automotive crash structures
Trilling, J., Schumacher, A., & Zhou, M. (2024) · Applied Intelligence · doi:10.1007/s10489-024-05276-6 · Cited by 10
A Recursively Recurrent Neural Network (R2N2) Architecture for Learning Iterative Algorithms
Doncevic, D. T., Mitsos, A., Guo, Y., Li, Q., Dietrich, F., Dahmen, M., & Kevrekidis, I. G. (2024) · SIAM Journal on Scientific Computing · doi:10.1137/22m1535310 · Cited by 16
Developing Rapport between Humans and Machines: Emotionally Intelligent AI Assistants
Agrawal, R., & Pandey, N. (2024) · International Journal for Research in Applied Science and Engineering Technology · doi:10.22214/ijraset.2024.59015 · Cited by 11
Tang, Y. F., Guo, S., Liu, J., Wan, B., An, L., & Liu, J. K. (2024) · Knowledge-Based Systems · doi:10.1016/j.knosys.2024.111736 · Cited by 10
Application of reinforcement learning to deduce nuclear power plant severe accident scenario
Song, S. H., Lee, Y., Bae, J. Y., Song, K. S., Seo, M. R., Kim, S. J., & Lee, J. I. (2024) · Annals of Nuclear Energy · doi:10.1016/j.anucene.2024.110605 · Cited by 11
Evolutionary reinforcement learning with action sequence search for imperfect information games
Wu, X., Zhu, Q., Chen, W., Lin, Q., Li, J., & Coello, C. A. C. (2024) · Information Sciences · doi:10.1016/j.ins.2024.120804 · Cited by 10
Metadata
- Authors
- Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I., Lai, M., Guez, A., Lanctot, M., Sifre, L., Kumaran, D., Graepel, T., Lillicrap, T., Simonyan, K., & Hassabis, D.
- Affiliations
- Google DeepMind (United Kingdom) (GB)
- Corresponding author
- Silver, David
- Hubert, Thomas
- Schrittwieser, Julian
- Hassabis, Demis
- ORCID
- Venue
- Science
- Published
- 2018-12-07
- Volume
- 362
- Issue
- 6419
- Pages
- 1140-1144
- ISSN
- 0036-8075 (ISSN-L)
- 1095-9203
- OpenAlex
- W2902907165
- Cited by
- 3,562
- References
- 30
- Citations by year
- 2026: 257 · 2025: 427 · 2024: 512 · 2023: 514 · 2022: 569 · 2021: 546 · 2020: 462 · 2019: 256 · 2018: 13
- FWCI
- 179.89
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Science (publishedVersion)
- PubMed (publishedVersion)
- UCL Discovery (University College London) (submittedVersion)
- Funders
- DeepMind
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- AlphaZero, chess, shogi, Go, evaluation function learning, game tree search
- MeSH terms
- Algorithms, Artificial Intelligence, Humans, Reinforcement, Psychology, Software, Video Games
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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.