Human-level control through deep reinforcement learning
The DOI matches its Crossref record; 38 of the 38 references it cites were found in a registry or an index.
Abstract
The theory of reinforcement learning provides a normative account, deeply rooted in psychological and neuroscientific perspectives on animal behaviour, of how agents may optimize their control of an environment. To use reinforcement learning successfully in situations approaching real-world complexity, however, agents are confronted with a difficult task: they must derive efficient representations of the environment from high-dimensional sensory inputs, and use these to generalize past experience to new situations. Remarkably, humans and other animals seem to solve this problem through a harmonious combination of reinforcement learning and hierarchical sensory processing systems, the former evidenced by a wealth of neural data revealing notable parallels between the phasic signals emitted by dopaminergic neurons and temporal difference reinforcement learning algorithms. While reinforcement learning agents have achieved some successes in a variety of domains, their applicability has previously been limited to domains in which useful features can be handcrafted, or to domains with fully observed, low-dimensional state spaces. Here we use recent advances in training deep neural networks to develop a novel artificial agent, termed a deep Q-network, that can learn successful policies directly from high-dimensional sensory inputs using end-to-end reinforcement learning. We tested this agent on the challenging domain of classic Atari 2600 games. We demonstrate that the deep Q-networ…
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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?
In the citing paper
“In previous studies, the agents are assumed to act based on simple and heuristic policy functions, and learned in isolated and restricted environments [1][2][3][4][5][6].”
From A survey on large language model based autonomous agents (2024) · cites this paper as [1]
In the citing paper
“This technique was developed in 2013 with Google Deep Mind [55].”
From Review of deep learning: concepts, CNN architectures, challenges, applications, future directions (2021) · cites this paper as [55]
In the citing paper
“Whereas previous reinforcement-learning algorithms often rely only on task-specific learning signals9,10, Dreamer rests predominantly on the unsupervised objective of its world model.”
From Mastering diverse control tasks through world models (2025) · cites this paper as [9]
In the citing paper
“Traditional RL such as AlphaGo Zero and the DQN series (Silver et al., 2017; Mnih et al., 2015; Yue et al., 2023) can continuously improve the performance of a policy in environments like Go and Atari games without an explicit upper bound .”
From Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model? (2025) · cites this paper as Mnih et al., 2015
In the citing paper
“Unlike previous reinforcement learning algorithms that often rely only on task-specific learning signals 7,8 , Dreamer rests predominantly on the unsupervised objective of its world model.”
From Mastering Diverse Domains through World Models (2023) · cites this paper as [7]
24 more claims about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 1×
“We use one particularly successful architecture, the deep convolutional network 17 , which uses hierarchical layers of tiled convolutional filters tomimictheeffectsofreceptivefields—inspiredbyHubelandWiesel’s seminalworkonfeedforwardprocessinginearlyvisualcortex 18 —thereby…”
Method(2008) · Journal of Machine Learning Research · Cited by 33,051
Reinforcement Learning: An Introduction
Sutton, R. S., & Barto, A. G. (1998) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.1998.712192 · Cited by 27,225
Reinforcement Learning: An Introduction
(2005) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.2004.842673 · Cited by 24,496
Reducing the dimensionality of data with neural networks
Hinton, G. E., & Salakhutdinov, R. R. (2006) · Science · doi:10.1126/science.1127647 · Cited by 20,928
(2010) · International Conference on Machine Learning · Cited by 13,995
A neural substrate of prediction and reward
Schultz, W., Dayan, P., & Montague, P. R. (1997) · Science · doi:10.1126/science.275.5306.1593 · Cited by 9,388 · cited 1×
“…4,5 , the former evidenced by a wealth of neural data revealingnotableparallelsbetweenthephasicsignalsemittedbydopa-minergicneuronsandtemporaldifferencereinforcementlearning algorithms 3 .”
BackgroundLearning Deep Architectures for AI
Bengio, Y. (2009) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000006 · Cited by 9,239 · cited 2×
“Notably, recent advances in deep neural networks 9–11 , in which several layers of nodes areused to buildup progressivelymore abstractrepresentations of the data, have made it possible for artificial neural networks to learn concepts such as object categories directly from raw sensory data.”
BackgroundWatkins, C. J. C. H., & Dayan, P. (1992) · Machine Learning · doi:10.1007/bf00992698 · Cited by 9,083
Cited by
Items 1,401–1,420 of 1,420
Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks
Matthews, M., Beukman, M., Lu, C., & Foerster, J. (2024) · arXiv · doi:10.48550/arxiv.2410.23208 · Cited by 1
“Through massive parallelisation and elimination of CPU-GPU transfer, this gives tremendous speed benefits (Lu et al., 2022).”
InfluentialComparing Task Simplifications to Learn Closed-Loop Object Picking Using Deep Reinforcement Learning
Breyer, M., Furrer, F., Novkovic, T., Siegwart, R., & Nieto, J. (2018) · arXiv · doi:10.48550/arxiv.1803.04996 · Cited by 1
Frank, G., Hota, G., Wang, K., Deng, C., Arora, K., Vins, D., Uppal, A., Olajide, O., Yoshimoto, K., Wang, Q., et al. (2026) · npj Unconventional Computing · doi:10.1038/s44335-026-00062-8
UBCL: A Reinforcement Learning Framework for Controllable and Diverse Player Behaviors
Cilan, A., & Ozgovde, A. (2026) · IEEE Transactions on Games · doi:10.1109/tg.2026.3703355
Serevan, B., & Kuşakçı, A. O. (2026) · International Journal of Lean Six Sigma · doi:10.1108/ijlss-02-2026-0094
Reinforcement Learning with History-Dependent Dynamic Contexts
Tennenholtz, G., Merlis, N., Shani, L., Mladenov, M., & Boutilier, C. (2023) · arXiv · doi:10.48550/arxiv.2302.02061
“RL has been applied to numerous domains, including recommender systems, robot control, video games, and autonomous vehicles (Afsar et al., 2022; Tessler et al., 2019; Mnih et al., 2015; Fayjie et al., 2018).”
MethodReVISE: Learning to Refine at Test-Time via Intrinsic Self-Verification
Lee, H., Oh, S., Kim, J., Shin, J., & Tack, J. (2025) · arXiv · doi:10.48550/arxiv.2502.14565 · Cited by 26
“However, RL is a complex and often unstable procedure (Mnih et al., 2015; Rafailov et al., 2023), and it does not explicitly model the verification of intermediate reasoning steps, making it difficult to assess whether a model is confident in its current trajectory or prone to deviating toward…”
Space-sampled Value Decay: Forgetting Mechanisms for Non-stationary Reinforcement Learning
Störck, F., Hartmann, P., Hinder, F., Neumann, K., & Hammer, B. (2026) · arXiv · doi:10.48550/arxiv.2606.11797
Ethics-Aware Safe Reinforcement Learning for Rare-Event Risk Control in Interactive Urban Driving
Li, D., & Okhrin, O. (2026) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2026.3708405
- Self-citation
NavFormer: A Transformer Architecture for Robot Target-Driven Navigation in Unknown and Dynamic Environments
Wang, H., Tan, A. H., & Nejat, G. (2024) · arXiv · Cited by 40
“Task-specific feature extraction has been shown to enhance policy learning in robot navigation [33].”
BackgroundUBCL: A Reinforcement Learning Framework for Controllable and Diverse Player Behaviors
Cilan, A., & Özgövde, A. (2025) · arXiv
Pritzel, A., Uria, B., Srinivasan, S., Puigdomènech, A., Vinyals, O., Hassabis, D., Wierstra, D., & Blundell, C. (2017) · arXiv · doi:10.48550/arxiv.1703.01988
Self-citationSequence Modeling of Temporal Credit Assignment for Episodic Reinforcement Learning
Liu, Y., Luo, Y., Zhong, Y., Chen, X., Liu, Q., & Peng, J. (2019) · arXiv · doi:10.48550/arxiv.1905.13420
Wang, R., Lehman, J., Clune, J., & Stanley, K. O. (2019) · arXiv
“Overall, ES has exhibited performance on par with some of the traditional, simple gradient-based RL algorithms on difficult RL domains (e.g. DQN [7] and A3C [60]), including Atari environments and simulated robot locomotion [47, 61].”
BackgroundAccelerating Reinforcement Learning through GPU Atari Emulation
Dalton, S., Frosio, I., & Garland, M. (2019) · arXiv · doi:10.48550/arxiv.1907.08467
Fast Task Inference with Variational Intrinsic Successor Features
Hansen, S., Dabney, W., Barreto, A., Van de Wiele, T., Warde-Farley, D., & Mnih, V. (2019) · arXiv · doi:10.48550/arxiv.1906.05030
Self-citationDecision Transformer: Reinforcement Learning via Sequence Modeling
Chen, L., Lu, K., Rajeswaran, A., Lee, K., Grover, A., Laskin, M., Abbeel, P., Srinivas, A., & Mordatch, I. (2021) · arXiv · doi:10.48550/arxiv.2106.01345
“(2020), representing 500 thousand of the 50 million transitions observed by an online DQN agent (Mnih et al., 2015) during training; we report the mean and standard deviation of 3 seeds.”
BackgroundHoffer, E., Hubara, I., & Soudry, D. (2017) · arXiv · doi:10.48550/arxiv.1705.08741
PettingZoo: Gym for Multi-Agent Reinforcement Learning
Terry, J. K., Black, B., Grammel, N., Jayakumar, M., Hari, A., Sullivan, R., Santos, L., Perez, R., Horsch, C., Dieffendahl, C., et al. (2020) · arXiv · doi:10.48550/arxiv.2009.14471
“the maximum expected discounted reward when playing in an environment. OpenAI Gym (Brockman et al., 2016) was introduced shortly after the potential of reinforcement learning became widely known with Mnih et al. (2015). At the time, doing basic research in reinforcement learning was a large engineering challenge. The…”
Background
Metadata
- Authors
- Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., Petersen, S., Beattie, C., Sadik, A., Antonoglou, I., King, H., Kumaran, D., Wierstra, D., Legg, S., & Hassabis, D.
- Affiliations
- Google DeepMind (United Kingdom) (GB)
- Corresponding author
- Kavukcuoglu, Koray
- Hassabis, Demis
- ORCID
- Venue
- Nature
- Published
- 2015-02-25
- Volume
- 518
- Issue
- 7540
- Pages
- 529-533
- ISSN
- 0028-0836 (ISSN-L)
- 1476-4687
- OpenAlex
- W2145339207
- Semantic Scholar
- 340f48901f72278f6bf78a04ee5b01df208cc508
- Cited by
- 30,357
- References
- 37
- Citations by year
- 2026: 2851 · 2025: 3619 · 2024: 3335 · 2023: 3592 · 2022: 3367 · 2021: 3971 · 2020: 3466 · 2019: 2907 · 2018: 1776 · 2017: 912 · 2016: 416 · 2015: 132
- FWCI
- 1057.57
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- www.springer.com/tdm (tdm)
- www.springer.com/tdm (vor)
- Locations
- Nature (publishedVersion)
- PubMed (publishedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- UCL Discovery (University College London) (submittedVersion)
- Funders
- DeepMind
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- deep Q-network, deep reinforcement learning, game-playing agents, end-to-end learning, deep neural network, temporal difference learning, dopaminergic neurons, human-level performance, DeepMind, AlphaGo, AlphaZero
- MeSH terms
- Algorithms, Artificial Intelligence, Humans, Models, Psychological, Reinforcement, Psychology, Reward, Neural Networks, Computer, 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.
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.