A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Open accessThe 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.
Abstract
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 1,901–2,000 of 3,633
INVIGORATE: Interactive Visual Grounding and Grasping in ClutterPreprint
Zhang, H., Lu, Y., Yu, C., Hsu, D., La, X., & Zheng, N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.11092 · Cited by 39
“To scale up complex decision making, both Silver et al. [32, 33, 34] and Cai et al. [2] use planning in the short term and use learning for long-term prediction.”
BackgroundContinuous Homeostatic Reinforcement Learning for Self-Regulated Autonomous AgentsPreprint
Laurençon, H., Segerie, C.-R., Lussange, J., & Gutkin, B. S. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.06580 · Cited by 9
Identifying Reasoning Flaws in Planning-Based RL Using Tree ExplanationsPreprint
Lam, K.-H., Lin, Z., Irvine, J., Dodge, J. E., Shureih, Z., Khanna, R., Kahng, M., & Fern, A. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.13978 · Cited by 5
“Our agent architecture is similar to AlphaZero [11], an RL agent based on game-tree search that demonstrated mastery of Go, Chess, and Shogi.”
MethodLarie, D., An, G. C., & Cockrell, C. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.02.17.480940 · Cited by 5
“Specifically, the latter project utilized the same method, DRL, as has been used by successful ML/Artificial Intelligence (AI) systems, the game-playing AIs and their descendants from Deep Mind (16–20).”
MethodReinforcement Learning in Quantitative Trading: A SurveyPreprint
Alameer, A., Saleh, H., & Alshehri, K. A. (2022) · doi:10.36227/techrxiv.19303853.v1 · Cited by 7
On Solving a Stochastic Shortest-Path Markov Decision Process as Probabilistic Inference
Baioumy, M., Lacerda, B., Duckworth, P., & Hawes, N. (2021) · Communications in Computer and Information Science · doi:10.1007/978-3-030-93736-2_58 · Cited by 4
Wilhelm, C., Steckelberg, A., & Rebitschek, F. G. (2024) · Systematic Reviews · doi:10.1186/s13643-024-02646-6 · Cited by 5
Data-Driven Reinforcement Learning for Mission Engineering and Combat Simulation
Henslee, A. C., Shukla, I., Dozier, H. R., Hansen, B., Arnold, T., Jabour, J. E., Thompson, B., Turner, G., White, J., & Dettwiller, I. (2024) · IUTAM bookseries · doi:10.1007/978-3-031-39303-7_21 · Cited by 3
Sun, C., Shen, S., Xue, D., Tao, W., & Zhou, Z. (2024) · IEEE Transactions on Games · doi:10.1109/tg.2024.3520970 · Cited by 6
“Self-play, a widely adopted training paradigm [7], [8], has shown promise in improving the strength of DRL-based AI-bots.”
Investigations into deep Reinforcement Learning for wind farm set-point optimisation
Sheehan, H., Poole, D. J., Filho, T. M. S., Bossanyi, E. A., & Landberg, L. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.127627 · Cited by 4
FrameGym: A reinforcement learning environments for steel frame structures
Du, M., Gao, Y., Wang, W., & Fu, B. (2025) · Engineering Structures · doi:10.1016/j.engstruct.2025.120991 · Cited by 5
Conformity towards humans versus AI in different task domains: the type of task matters
Thorp, S. O., Slupphaug, K. S., Rimol, L. M., Lervik, S., Hoel, S. K. B., & Grassini, S. (2025) · Journal of Psychology and AI · doi:10.1080/29974100.2025.2540762 · Cited by 5
Engineering Application Progress of Multi-Agent Deep Reinforcement Learning
Bao, K.-H., Shi, T., & Wang, S. J. (2025) · International Conference on Control and Automation · doi:10.1109/icca65672.2025.11129859 · Cited by 3
Parnada, A., Qu, M., Castellani, M., Chang, H. J., & Wang, Y. (2025) · Proceedings of the Institution of Mechanical Engineers Part I Journal of Systems and Control Engineering · doi:10.1177/09596518251350353 · Cited by 8
“Recent works show the success of RL-trained deep Artificial Neural Networks (ANNs) applied to many other areas, such as playing games 11–14 autonomous ground 15,16 and air navigation, 17,18 robotic manipulation 19–22 and recommender systems.”
Villarrubia, J., Costero, L., Igual, F. D., & Olcoz, K. (2025) · Future Generation Computer Systems · doi:10.1016/j.future.2025.108145 · Cited by 3
Energy-Saving Full-Body Control of Hybrid Vehicles Through Deep Reinforcement Learning
Chen, J., Tang, X., & Qin, Y. (2025) · IEEE Transactions on Transportation Electrification · doi:10.1109/tte.2025.3641893 · Cited by 3
“Now, various fields that have achieved considerable effects include games [14], robotics [15], autopilot [16], and academic studies [17].”
Minor embedding for quantum annealing with reinforcement learning
Nembrini, R., Dacrema, M. F., & Cremonesi, P. (2026) · Quantum Machine Intelligence · doi:10.1007/s42484-026-00341-4 · Cited by 6
“This results in S H being the same mask applied via IAM (see Section 2.4.3) to the policy, as shown in Figure 5.”
InfluentialAlleviating Local Optima in Bridge Bidding via Diverse PPO Ensembling
Zhang, Y., Song, F., Zhang, C., Zhang, T., & Deng, W. (2025) · IEEE Access · doi:10.1109/access.2025.3555303 · Cited by 3
Zhang, M., Yousif, M. Z., Xu, M., Zhou, H., Yu, L., & Lim, H.-C. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.112468 · Cited by 7
“Deep Reinforcement Learning (DRL), which is a ML method where an agent learns to make decisions by interacting with an environment, has shown remarkable results in fields such as robotics [32], game-playing [33, 34], and optimisation problems [35, 36].”
A Hardware‐Aware Photonic Spiking‐DDPG Reinforcement Learning Architecture for Continuous Control
Zeng, X., Xiang, S., Zhao, H., Han, Y., Yu, W., Huang, Z., Shi, S., Guo, X., Zhang, Y., Shi, Y., et al. (2026) · Laser & Photonics Review · doi:10.1002/lpor.202502481 · Cited by 3
“A representative example is AlphaGo, which leveraged RL principles to defeat human professional players for the first time [ 22, 23 ].”
Fink, O., Nejjar, I., Sharma, V., Niresi, K. F., Sun, H., Dong, H., Xu, C., Wei, A., Bizzi, A. A. C., Theiler, R., et al. (2026) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2026.112376 · Cited by 26
Bouguettaya, S., Zedadra, O., Pupo, F., & Fortino, G. (2026) · Artificial Intelligence Science and Engineering · doi:10.23919/aise.2026.000001 · Cited by 4
Waldrop, M. M. (2022) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2214148119 · Cited by 4
“Since then, DeepMind has continued to refine and generalize this approach (2, 3).”
BackgroundA statistical approach for detecting AI-assisted cheating in the game of Go
Park, J., Im, J., On, S., Lee, S. J., & Lee, J. (2022) · Journal of the Korean Physical Society · doi:10.1007/s40042-022-00622-8 · Cited by 3
Hidden Information General Game Playing with Deep Learning and Search
Partridge, Z., & Thielscher, M. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20868-3_12 · Cited by 3
A Minimal “Functionally Sentient” Organism Trained With Backpropagation Through Time
Var, M. P., Fairbank, M., & Samothrakis, S. (2023) · Adaptive Behavior · doi:10.1177/10597123231166416 · Cited by 3
“Existing Reinforcement Learning (RL) and Adaptive Dynamic Programming (ADP) algorithms have become more ef fi cient in examining, learning and solving new problems (e.g., a new game) (Mnih et al., 2013; Silver et al., 2018).”
BackgroundAdaptive Vibration Control of Smart Structure Using Deep Reinforcement Learning
Honda, S., Imura, Y., SASAKI, K., & Takeda, R. (2022) · EPI International Journal of Engineering · doi:10.25042/epi-ije.082022.03 · Cited by 3
“In particular, deep reinforcement learning, which utilizes deep learning for conventional reinforcement learning, is characterized by its ability to handle complex problems with nonlinearities and has demonstrated performance that can beat human players in the field of games such as chess [6].”
BackgroundInverted Pendulum Control: A Comparative Study from Conventional Control to Reinforcement Learning
Ataka, A., Sandiwan, A. P., Tnunay, H., Utomo, D. R., & Cahyadi, A. I. (2023) · Jurnal Nasional Teknik Elektro dan Teknologi Informasi (JNTETI) · doi:10.22146/jnteti.v12i3.7065 · Cited by 3
“Since its huge success in solving games called Atari [3], [4], deep reinforcement learning has been applied successfully to solving various games such as chess, Go, and Dota2 [5]–[7].”
BackgroundCenteno, T. B., Ferreira, C. A., Guillen, J. G. I., Echeverry, A. V., Buendia, R. H., Alanya, O. D. V., Moya, S. M., Aliaga, D. C. R., Centeno, W. G., Miranda, B. S. A., et al. (2023) · Revista de Biología Tropical · doi:10.15517/rev.biol.trop..v71i1.51310 · Cited by 3
PathletRL: Trajectory Pathlet Dictionary Construction using Reinforcement Learning
Alix, G., & Papagelis, M. (2023) · ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL) · doi:10.1145/3589132.3625622 · Cited by 3
“RL methods have seen success in solving decision-based problems in an attempt to maximize rewards [28, 44].”
MethodAdaptive Virtual Assistant for Virtual Reality-based Remote Learning
Sloan, H., Zhao, R., Aqlan, F., Yang, H. Y., & Zhu, R. (2024) · ASEE Annual Conference & Exposition · doi:10.18260/1-2--41234 · Cited by 4
Interpretable and Predictive Models Based on High‐Dimensional Data in Ecology and Evolution
Jahner, J. P., Buerkle, C. A., Gannon, D. G., Grames, E. M., McFarlane, S. E., Siefert, A., Bell, K. L., DeLeo, V. L., Forister, M. L., Harrison, J. G., et al. (2026) · Molecular Ecology Resources · doi:10.1111/1755-0998.70191 · Cited by 3
“Some of the most remarkable models in society, such as those for large language modelling (Zhao et al. 2026), natural voice recognition (Xiong et al. 2016), image segmentation (Kirillov et al. 2023), and board game algorithms (Silver et al. 2018), are typically trained on enormous sample sizes.”
Cotogni, M., & Cusano, C. (2024) · Pattern Recognition Letters · doi:10.1016/j.patrec.2024.05.013 · Cited by 4
BrainQN: Enhancing the Robustness of Deep Reinforcement Learning with Spiking Neural Networks
Feng, S., Cao, J. W., Ou, Z., Chen, G., Zhong, Y., Wang, Z., Yan, J., Chen, J., Wang, B., Zou, C., et al. (2024) · Advanced Intelligent Systems · doi:10.1002/aisy.202400075 · Cited by 5
Policy Testing with MDPFuzz (Replicability Study)
Mazouni, Q., Spieker, H., Arnaud Gotlieb, B., & Acher, M. (2024) · ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA) · doi:10.1145/3650212.3680382 · Cited by 5
“In recent years, the combination of Machine Learning (ML) with neural networks has been effectively employed to solve complex sequential decision-making tasks of diverse nature, such as gaming [24], planning [12, 30] or system control [4, 19].”
The phenomenon of artificial intelligence in the context of the evolution of existence
Maltsev, Y. V. (2024) · Философская мысль · doi:10.25136/2409-8728.2024.7.70674 · Cited by 3
A concise review of intelligent game agent
Li, H., Pang, X., Sun, B., & Liu, K. (2024) · Entertainment Computing · doi:10.1016/j.entcom.2024.100894 · Cited by 9
Ren, Z., Zhu, Y., Feng, Z., Wan, H., Xu, Y., Lin, J., & Wu, Z. (2024) · Journal of Industrial and Management Optimization · doi:10.3934/jimo.2024160 · Cited by 3
Artificial Intelligence-Enabled Innovative and Advanced Solutions for Banking and Finance Sector
Arumugam, S. K., Susendiran, S., Seranmadevi, R., et al. (2024) · Advances in logistics, operations, and management science book series · doi:10.4018/979-8-3693-6200-6.ch007 · Cited by 3
A Parallel Framework for Fast Charge/Discharge Scheduling of Battery Storage Systems in Microgrids
Huang, W., Chung, W., Wu, C., et al. (2024) · Energies · doi:10.3390/en17246371 · Cited by 3
Combining simulation and reinforcement learning to reduce food waste in food retail
Pilarski, S., Sidhu, A., & Varró, D. (2024) · SIMULATION · doi:10.1177/00375497241299054 · Cited by 5
Empowering branch-and-bound algorithms via reinforcement and imitation learning for enhanced search
Wang, Q. (2025) · Applied Soft Computing · doi:10.1016/j.asoc.2025.112690 · Cited by 4
Njiru, R. D. G., Dong, C., Appel, F., & Balmann, A. (2025) · The International Food and Agribusiness Management Review · doi:10.22434/ifamr.1126 · Cited by 3
Deep Reinforcement Learning for Optimizing Route Planning in Urban Traffic
Mittal, M., Sehgal, A., Varshney, N., et al. (2025) · doi:10.1109/icccit62592.2025.10928029 · Cited by 3
“RL with deep neural networks (DNNs) is known as DRL because DRL integrates with RL to handle largescale environments and approximate complicated functions [24].”
A novel approach to feedback control with deep reinforcement learning
Bidi, K. A., Coron, J., Hayat, A., & Lichtlé, N. (2025) · Systems & Control Letters · doi:10.1016/j.sysconle.2025.106102 · Cited by 3
“In the realm of games, RL algorithms have achieved superhuman performance in challenging domains like Go, Chess, StarCraft [13,14], as well as in classical Atari games [15].”
Online Decision-Making Under Uncertainty for Vehicle-to-Building Systems
Sen, R., Zhang, Y., Liu, F., Talusan, J. P., Pettet, G., Suzue, Y., Mukhopadhyay, A., & Dubey, A. K. (2025) · ACM/IEEE International Conference on Cyber-Physical Systems (ICCPS) · doi:10.1145/3716550.3722024 · Cited by 6
“We collected real-world data from NATC-SV in Santa Clara, California, USA, covering building power consumption, EV charger usage, and EV telemetry over a nine-month period from May 2023 to January 2024.”
Gong, M., Yi, Z., Chen, Y., Li, H., & Cui, Y. (2025) · Applied Intelligence · doi:10.1007/s10489-025-06693-x · Cited by 5
Hu, Z., Liang, X., Zhang, J., Xiaochuan, Y. O. U., & Chengcheng, M. A. (2025) · Chinese Journal of Aeronautics · doi:10.1016/j.cja.2025.103663 · Cited by 5
Deep Reinforcement Learning-Based Deployment Method for Emergency Communication Network
Huang, B., Lu, Y., Ma, H., Yin, C., Yang, R., Shi, Y., Tao, Y., Wen, Y., & Zhong, Y. (2025) · Applied Sciences · doi:10.3390/app15147961 · Cited by 3
CausalCOMRL: Context-based offline meta-reinforcement learning with causal representation
Zhang, Z., Meng, W., Sun, H., & Pan, G. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.107955 · Cited by 9
A study of a matrix manufacturing system scheduling method considering equipment degradation
Pei, F., Xiang, H., Zhuang, C., et al. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.129363 · Cited by 3
Chaos-based reinforcement learning with TD3
Matsuki, T., Sakemi, Y., & Aihara, K. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.108202 · Cited by 9
“…on deep reinforcement learning, which incorporates deep learning techniques, has become popular, and this approach has made it possible to learn even more difficult tasks [19, 20] Deep reinforcement learning can also provide effective learning performance in more complex tasks such as Go [21, 22].”
Impartial Games: A Challenge for Reinforcement Learning
Zhou, B., & Riis, S. (2026) · Machine Learning · doi:10.1007/s10994-026-06996-1 · Cited by 12
“The advent of AlphaZero and its variations has marked a significant success in machine learning, demonstrating superhuman capabilities across complex strategy games such as chess, Go, and shogi (Silver et al., 2017, 2018; Schrittwieser et al., 2020; Sadler and Regan, 2019).”
InfluentialAljaafreh, A., & Al-Oudat, N. (2019) · Procedia Computer Science · doi:10.1016/j.procs.2019.09.463 · Cited by 3
Shalosh B. Ekhad: a computer credit for mathematicians
Zhi, H., Cui, Y., Zhang, J., & Eviston-Putsch, J. (2019) · Scientometrics · doi:10.1007/s11192-019-03305-7 · Cited by 4
Experimental semi-autonomous eigensolver using reinforcement learning
Pan, C.-Y., Hao, M., Barraza, N., Solano, E., & Albarrán-Arriagada, F. (2021) · Scientific Reports · doi:10.1038/s41598-021-90534-7 · Cited by 3
“(20) |E � = 1 √ 2 (|00� − |11�), α = −1, |E ((1))� = 1 √ 2 (−|01� + |10�), α((1)) = −1, |E ((2))� = 1 √ 2 (|00� + |11�), α((2)) = 1, |E ((3))� = 1 √ 2 (|01� + |10�), α((3)) = 1.”
Schaeffer, J. (2020) · ICGA Journal · doi:10.3233/icg-200149 · Cited by 3
The power of randomization by sex in multilocus genetic evolution
Vasylenko, L., Feldman, M. W., & Livnat, A. (2020) · Biology Direct · doi:10.1186/s13062-020-00277-0 · Cited by 3
Cohen-Solal, Q., & Cazenave, T. (2023) · Adaptive Agents and Multi-Agents Systems · doi:10.65109/feki3957 · Cited by 18
Cao, Y., & Zhan, H. (2021) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.12270 · Cited by 9
Predictability of AI Decisions
Musiolik, G. (2020) · Advances in computational intelligence and robotics book series · doi:10.4018/978-1-7998-3499-1.ch002 · Cited by 6
Learning Camera Control in Dynamic Scenes from Limited Demonstrations
Hanocka, R., Assa, J., Cohen‐Or, D., & Giryes, R. (2022) · Computer Graphics Forum · doi:10.1111/cgf.14444 · Cited by 3
Conversational Co-creativity with Deep Reinforcement Learning Agent in Kitchen Layout
Hsieh, P., Benrós, D., & Dogan, T. (2022) · Springer eBooks · doi:10.1007/978-3-030-90625-2_23 · Cited by 3
Modular Construction Planning Using Graph Neural Network Heuristic Search
Hawkins, P., Maire, F., Denman, S., & Baktashmotlagh, M. (2022) · Lecture notes in computer science · doi:10.1007/978-3-030-97546-3_19 · Cited by 5
What the Near Future of Artificial Intelligence Could Be
Floridi, L. (2021) · Philosophical studies series · doi:10.1007/978-3-030-81907-1_22 · Cited by 4
Solving problems of the oil and gas sector using machine learning algorithms
(2021) · Acta Montanistica Slovaca · doi:10.46544/ams.v26i2.11 · Cited by 4
“It considers all possible actions but explores the most promising ones (Silver et al., 2018).”
BackgroundA Comparative Tutorial of Bayesian Sequential Design and Reinforcement Learning
Tec, M., Duan, Y., & Müller, P. (2022) · The American Statistician · doi:10.1080/00031305.2022.2129787 · Cited by 8
“…example, popular successful RL applications with simulation-based environments include Atari video-games (Mnih et al., 2013), complex board games like chess and Go (Silver et al., 2018), robotic tasks (Tunyasuvunakool et al., 2020) and autonomous driving (Sallab et al., 2017; Wurman et al., 2022).”
BackgroundCombining Monte-Carlo Tree Search with Proof-Number Search
Doe, E., Winands, M. H. M., Soemers, D. J. N. J., et al. (2022) · IEEE Conference on Games (CoG) · doi:10.1109/cog51982.2022.9893635 · Cited by 3
“Shogi endgames have been one of the main test domains of PNS [31], whereas MCTS has become the dominating search technique for this game [32].”
BackgroundUnsupervised Hebbian Learning on Point Sets in StarCraft II
Kang, B., Kumar, H., Dash, S., & Mukhopadhyay, S. (2022) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn55064.2022.9892259 · Cited by 5
“Atari [1], Chess [2], and Go [3] games are well known examples where game AI demonstrates successful progress.”
BackgroundGMOM: An Offloading Method of Dependent Tasks Based on Deep Reinforcement Learning
Zheng, H., Yu, S., & Cui, X. (2022) · Mobile Information Systems · doi:10.1155/2022/9587040 · Cited by 3
“Boasting the strengths of both deep learning and reinforcement learning, DRL has strong perception and decision-making ability, so it has been widely studied and applied [11] to complex decision-making problems with high-dimensional state/action spaces, such as games [12] and robots [13].”
BackgroundSearch and learning for unsupervised text generation
Mou, L. (2022) · AI Magazine · doi:10.1002/aaai.12068 · Cited by 5
Vloga umetne inteligence v izobraževanju in za izobraževanje
Pesek, I., & Krašna, M. (2022) · doi:10.18690/um.ff.11.2022.13 · Cited by 15
Active Grammatical Inference for Non-Markovian Planning
Topper, N., Atia, G., Trivedi, A., & Velasquez, A. (2022) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v32i1.19853 · Cited by 5
“This assumption holds true for the impressive AlphaGo (Silver et al. 2016), AlphaGo Zero (Silver et al. 2017), and Al-phaZero (Silver et al. 2018) planning agents, but it is often the case that the underlying goal of an agent is non-Markovian.”
BackgroundEfficient Hierarchical Exploration with An Active Subgoal Generation Strategy
Xu, X., Zuo, G., Li, J., & Huang, G. M. (2022) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio55434.2022.10011930 · Cited by 3
“, board games [1],[2] and robotic control tasks [3],[4],[5].”
BackgroundGumbel MuZero for the Game of 2048
Kao, C.-Y., Guei, H., Wu, T.-R., & Wu, I. (2022) · Conference on Technologies and Applications of Artificial Intelligence (TAAI) · doi:10.1109/taai57707.2022.00017 · Cited by 3
“MuZero [7] is a generalized version of AlphaZero [6].”
BackgroundThe Role of a Reward in Shaping Multiple Football Agents’ Behavior: An Empirical Study
Kim, S. H., Kim, J., & Lee, J. (2023) · Applied Sciences · doi:10.3390/app13063622 · Cited by 3
Evolving Neuromorphic Systems on the Ethereum Smart Contract Platform
Wu, H., Fang, B., Xiang, C., Cohen, G., André, V. S., & Ramesh, B. (2022) · IEEE Global Emerging Technology Blockchain Forum: Blockchain & Beyond (iGETblockchain) · doi:10.1109/igetblockchain56591.2022.10087086 · Cited by 3
“The AI research field has individually demonstrated many significant and valuable concepts, evidenced by the most popular breakthrough with the release of alphaGo [2] which beat the leading world champion in the game of Go.”
BackgroundDataset Related Experimental Investigation of Chess Position Evaluation Using a Deep Neural Network
Wieczerzak, D., & Czarnul, P. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-30442-2_32 · Cited by 3
Xiang, W. (2023) · Socio-Ecological Practice Research · doi:10.1007/s42532-023-00153-x · Cited by 4
“…learning, AlphaZero learns how to play a game from scratch by playing the game against itself with no embedded human strategies but only the basic rules of the game (Campbell 2018; Kasparov 2018; McGrath et al. 2022, p. 1 of 10, p. 9 of 10; Silver et al. 2018a, p. 1140; Silver et al. 2018b).”
BackgroundInfluentialLocal Explanations for Reinforcement Learning
Luss, R., Dhurandhar, A., & Liu, M. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i7.26081 · Cited by 7
“Deep reinforcement learning (RL) has seen stupendous success over the last decade with superhuman performance in games such as Go (Silver, Huang, and et al. 2016), Chess (Silver et al. 2018), and Atari benchmarks (Mnih, Kavukcuoglu, and et al. 2015).”
BackgroundMarkolf, L., & Stursberg, O. (2023) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc55779.2023.10156344 · Cited by 3
“Deep learning [1] has significantly contributed to the success of machine learning in recent years, as reported, e.g., in [2] and [3].”
BackgroundDiscovering Editing Rules by Deep Reinforcement Learning
Mei, Y., Song, S., Fang, C., Wei, Z., Fang, J., & Long, J. (2023) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde55515.2023.00034 · Cited by 3
Automating Rigid Origami Design
Geiger, J., Martinkus, K., Richter, O., & Wattenhofer, R. P. (2023) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2023/645 · Cited by 4
New Hybrid Graph Convolution Neural Network with Applications in Game Strategy
Xu, H., Seng, K. P., & Ang, L.-M. (2023) · Electronics · doi:10.3390/electronics12194020 · Cited by 3
Cardsformer: Grounding Language to Learn a Generalizable Policy in Hearthstone
Xia, W., Yang, Y., Ruan, J., Xing, D., & Xu, B. (2023) · Frontiers in artificial intelligence and applications · doi:10.3233/faia230581 · Cited by 5
Cui, X., Yu, L., Xie, Z. Y., Liu, H., Yang, S., & Mou, C. (2023) · Electronics · doi:10.3390/electronics12194077 · Cited by 4
RAP Vol: Robust Adversary Populations With Volume Diversity Measure
Yang, J., & Zhang, J. (2023) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3317145 · Cited by 3
“The most representative achievements are the DRL models for board games, such as AlphaGo [2] and AlphaZero [3], which defeat the top human players.”
Playing Various Strategies in Dominion with Deep Reinforcement Learning
Gerigk, J., & Engels, S. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · doi:10.1609/aiide.v19i1.27518 · Cited by 3
The Path to AGI Goes through Embodiment
Tan, C., & Jaiswal, S. (2023) · Proceedings of the AAAI Symposium Series · doi:10.1609/aaaiss.v1i1.27485 · Cited by 6
“One simple example is chess, where rule-based AI sur-passed humans in 1997 (Campbell, Hoane Jr, and Hsu 2002), eventually followed by RL-based AI methods (Silver et al. 2018).”
MethodModel-based optimal action selection for Dyna-Q reverberation suppression cognitive sonar
Fu, Y., Ma, X., Feng, C. M., et al. (2023) · EURASIP Journal on Advances in Signal Processing · doi:10.1186/s13634-023-01054-7 · Cited by 3
“In the field of Go [11], to acquire decision-making capabilities by learning previous game actions and using them to subsequent actions.”
BackgroundMecanum-Wheeled Robot Control Based on Deep Reinforcement Learning
Setiadilaga, O., Cahyadi, A. I., & Ataka, A. (2023) · doi:10.1109/icitee59582.2023.10317659 · Cited by 7
“In recent years, deep reinforcement learning, which combines deep neural network and reinforcement learning framework, has been shown to perform successfully in various problems, such as Atari games [6], [7], chess, Go, as well as Dota2 [8]- [9].”
BackgroundChessEye: An Integrated Framework for Accurate and Efficient Chessboard Reconstruction
Ranasinghe, P., Ranasinghe, P., & Ashan, V. (2023) · Moratuwa Engineering Research Conference (MERCon) · doi:10.1109/mercon60487.2023.10355515 · Cited by 3
“In contrast, chess programs can perform these calculations within seconds [4].”
BackgroundAthénan wins sixteen gold medals at the Computer Olympiad
Cohen-Solal, Q., & Cazenave, T. (2024) · ICGA Journal · doi:10.3233/icg-230239 · Cited by 3
“Unlike Alpha Zero-like algorithms (Silver et al., 2018), Athénan is based on the Descent framework (Cohen-Solal, 2020).”
MethodRLOP: A Framework Design for Offset Prefetching Combined with Reinforcement Learning
Huang, Y., & Wang, Z. (2024) · Lecture notes in electrical engineering · doi:10.1007/978-981-99-9247-8_10 · Cited by 4
Improvement of move naturalness for playing good-quality games with middle-level players
Hsueh, C.-H., & Ikeda, K. (2024) · Applied Intelligence · doi:10.1007/s10489-023-05210-2 · Cited by 5
Designing Biological Sequences without Prior Knowledge Using Evolutionary Reinforcement Learning
Zeng, X., Hao, X., Tang, H., Tang, Z., Jiao, S., Lu, D., & Peng, J. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i1.27792 · Cited by 5
“Reinforcement learning (RL) can start without prior knowledge (Schrittwieser et al. 2020; Silver et al. 2018).”
BackgroundEntertainment Computing: Past, Present, and Future
Ciancarini, P., Nakatsu, R., & Rauterberg, M. (2024) · Springer eBooks · doi:10.1007/978-3-319-27648-9_34-1 · Cited by 4
The Ethics of Life as It Could Be: Do We Have Moral Obligations to Artificial Life?
Witkowski, O., & Schwitzgebel, E. (2024) · Artificial Life · doi:10.1162/artl_a_00436 · Cited by 8
“Current AI systems outperform even expert humans in games, such as chess, Go, and poker (N. Brown & Sandholm, 2019; Campbell et al., 2002; Silver et al., 2018; Silver et al., 2017), and in practical domains, such as lung cancer screening (Ardila et al., 2019), predicting protein structure (Jumper…”
BackgroundTemporally extended successor feature neural episodic control
Zhu, X. (2024) · Scientific Reports · doi:10.1038/s41598-024-65687-w · Cited by 3
DanZero+: Dominating the GuanDan Game Through Reinforcement Learning
Zhao, Y., Lu, Y., Zhao, J., Zhou, W., & Li, H. (2024) · IEEE Transactions on Games · doi:10.1109/tg.2024.3422396 · Cited by 3
“In this work, we aim to develop an AI program for an exceptionally complex and popular card game called GuanDan .”
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.