Deep Reinforcement Learning: A Brief Survey
Open accessThe DOI matches its Crossref record; 177 of the 182 references it cites were found in a registry or an index.
Abstract
Deep reinforcement learning (DRL) is poised to revolutionize the field of artificial intelligence (AI) and represents a step toward building autonomous systems with a higher-level understanding of the visual world. Currently, deep learning is enabling reinforcement learning (RL) to scale to problems that were previously intractable, such as learning to play video games directly from pixels. DRL algorithms are also applied to robotics, allowing control policies for robots to be learned directly from camera inputs in the real world. In this survey, we begin with an introduction to the general field of RL, then progress to the main streams of value-based and policy-based methods. Our survey will cover central algorithms in deep RL, including the deep Q-network (DQN), trust region policy optimization (TRPO), and asynchronous advantage actor critic. In parallel, we highlight the unique advantages of deep neural networks, focusing on visual understanding via RL. To conclude, we describe several current areas of research within the field.
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
“Foundational surveys on RL [7,72] provide broad overviews of classical and deep RL, respectively, but do not focus on human feedback or LLMs.”
From RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs (2026) · cites this paper as [7]
In the citing paper
“Classical approaches employ reinforcement learning (RL) [6, 7] and imitation learning [8-10] that operate on primitive actions, which could be challenging for systematic exploration [11-15], interpretability [16-18], and generalization [19-21].”
From Voyager: An Open-Ended Embodied Agent with Large Language Models (2023) · cites this paper as [7]
In the citing paper
“This can be frequently modeled as decision making and learning in a Markov decision process (MDP) [160].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [160]
Citations
Cites
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Learning representations by back-propagating errors
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986) · Nature · doi:10.1038/323533a0 · Cited by 31,892 · cited 2×
“The workhorse of DRL, however, remains backpropagation [162, 111].”
Human-level control through deep reinforcement learning
Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., et al. (2015) · Nature · doi:10.1038/nature14236 · Cited by 30,357 · cited 10×
“[34] train locally linear models for use with the NAF algorithm— the continuous equivalent of the DQN [71]—to improve the algorithm’s sample complexity in the robotic domain where”
InfluentialReinforcement 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 · cited 7×
“(temporal) credit assignment problem [115].”
InfluentialReinforcement Learning: An Introduction
(2005) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.2004.842673 · Cited by 24,496
Auto-Encoding Variational Bayes
Kingma, D. P., & Welling, M. (2013) · arXiv · doi:10.48550/arxiv.1312.6114 · Cited by 20,067 · cited 1×
“[22] devised a method for calculating gradients to optimize stochastic policies by “reparameterizing” [30], [60] the stochasticity away from the network, thereby allowing standard gradients to be used (instead of the high-variance REINFORCE estimator [97]).”
Labaca-Castro, R. (2023) · Machine Learning under Malware Attack · doi:10.1007/978-3-658-40442-0_9 · Cited by 19,272
GAN(Generative Adversarial Nets)
(2017) · Journal of Japan Society for Fuzzy Theory and Intelligent Informatics · doi:10.3156/jsoft.29.5_177_2 · Cited by 19,270
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 · cited 7×
“dynamic programming over all possible actions (Figure 3, top left), sample trajectories for heuristic search (as was done by AlphaGo [108]), or even perform an exhaustive search (Figure 3, top right).”
InfluentialDistilling the Knowledge in a Neural Network
Hinton, G., Vinyals, O., & Dean, J. (2015) · arXiv · doi:10.48550/arxiv.1503.02531 · Cited by 14,204 · cited 1×
“technique known as distillation [40] has been utilised to both transfer policies learned by large DQNs to smaller DQNs, and transfer policies learned by several DQNs trained on separate games to one single DQN [85, 95].”
Cited by
Items 601–700 of 1,021
Yu, T.-H., Su, B., Battalora, L. C., Liu, S., & Tseng, Y. J. (2021) · Briefings in Bioinformatics · doi:10.1093/bib/bbab377 · Cited by 64
“Other than ML, DL has been shown to produce incredible results in computer vision [29], speech recognition [30], reinforcement learning [31], text analysis [32] and even CNS drug research [33].”
BackgroundDdpg-Based Adaptive Voltage Control of Distribution Networks with Multi-Terminal Sop
Li, P., Wei, M., Ji, H., Xi, W., Yu, H., Wu, J., Yao, H., & Chen, J. (2021) · International Journal of Electrical Power & Energy Systems · doi:10.1016/j.ijepes.2022.108138 · Cited by 64
A survey on deep reinforcement learning architectures, applications and emerging trends
Balhara, S., Gupta, N., Alkhayyat, A. H., Bharti, I., Malik, R. Q., Mahmood, S. N., & Abedi, F. (2022) · IET Communications · doi:10.1049/cmu2.12447 · Cited by 73
Yu, P., Zhang, H., Song, Y., Hui, H., & Chen, G. (2023) · IEEE Transactions on Power Systems · doi:10.1109/tpwrs.2023.3237888 · Cited by 69
Chung, J., Fayyad, J., Younes, Y. A., & Najjaran, H. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-023-10670-6 · Cited by 79
“Combined together, DRL has gotten rapid traction and is currently thriving with accomplishments mastering the complexity of MAPF by employing neural networks as function approximators (Arulkumaran et al. 2017; Gronauer and Diepold 2022).”
Geske, A. M., Herold, D. M., & Kummer, S. (2024) · Journal of the Air Transport Research Society · doi:10.1016/j.jatrs.2024.100030 · Cited by 71
Behzadan, V., & Munir, A. (2021) · IEEE Intelligent Transportation Systems Magazine · doi:10.1109/mits.2019.2898964 · Cited by 58
Huang, Q., & Song, W. (2019) · Computers Environment and Urban Systems · doi:10.1016/j.compenvurbsys.2019.101360 · Cited by 63
SR360: boosting 360-degree video streaming with super-resolution
Chen, J., Hu, M., Luo, Z., Wang, Z., & Wu, D. (2020) · International Workshop on Network and Operating Systems Support for Digital Audio and Video (NOSSDAV) · doi:10.1145/3386290.3396929 · Cited by 63
A Survey on Applications of Reinforcement Learning in Flying Ad-Hoc Networks
Rezwan, S., & Choi, W. (2021) · Electronics · doi:10.3390/electronics10040449 · Cited by 73
RLOps: Development Life-Cycle of Reinforcement Learning Aided Open RAN
Li, P., Thomas, J. D., Wang, X., Khalil, A., Ahmad, A., Inacio, R., Kapoor, S., Parekh, A., Doufexi, A., Shojaeifard, A., et al. (2022) · IEEE Access · doi:10.1109/access.2022.3217511 · Cited by 79
“Development Operations Safety/Security MDP formulation [45] Metrics design [46] Algorithm design [47] Training methodologies [48] Explainability [49] Digital twins [50], [51] Sim2Real [52], [53] [54] Performance evaluation [55] A/B deployment [56] Model decay [57] Interoperability [58] Deployment…”
MethodTowards Better Interpretability in Deep Q-Networks
Annasamy, R. M., & Sycara, K. P. (2019) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v33i01.33014561 · Cited by 33
Emergent behaviour and neural dynamics in artificial agents tracking odour plumes
Singh, S. H., van Breugel, F., Rao, R. P. N., & Brunton, B. W. (2023) · Nature Machine Intelligence · doi:10.1038/s42256-022-00599-w · Cited by 61
A novel multi-step Q-learning method to improve data efficiency for deep reinforcement learning
Yuan, Y., Yu, Z. L., Gu, Z., Yeboah, Y., Wu, W., Deng, X., Li, J., & Li, Y. (2019) · Knowledge-Based Systems · doi:10.1016/j.knosys.2019.03.018 · Cited by 68
A Survey of Artificial Intelligence in Cybersecurity
Morovat, K., & Panda, B. (2020) · International Conference on Computational Science and Computational Intelligence (CSCI) · doi:10.1109/csci51800.2020.00026 · Cited by 76
“…E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.” Generally, there are three learning algorithms for training machines, as defined below: Supervised learning : where data is limited or not given [16].”
BackgroundDeep Reinforcement Learning for Cybersecurity Threat Detection and Protection: A Review
Sewak, M., Sahay, S. K., & Rathore, H. (2022) · Communications in computer and information science · doi:10.1007/978-3-030-97532-6_4 · Cited by 84
“This is where the distinction between RL (also known as Classical RL) and Deep Reinforcement Learning (DRL) [4,5] comes in.”
BackgroundA deep reinforcement transfer convolutional neural network for rolling bearing fault diagnosis
Wu, Z., Jiang, H., Liu, S., & Wang, R. (2022) · ISA Transactions · doi:10.1016/j.isatra.2022.02.032 · Cited by 61
Ochoa, T., Gil, E., Angulo, A., & Valle, C. (2022) · Applied Energy · doi:10.1016/j.apenergy.2022.119067 · Cited by 65
Multi-type data fusion framework based on deep reinforcement learning for algorithmic trading
Liu, P., Zhang, Y., Bao, F., Yao, X., & Zhang, C. (2022) · Applied Intelligence · doi:10.1007/s10489-022-03321-w · Cited by 61
A review on interaction control for contact robots through intent detection
Li, Y., Sena, A., Wang, Z., Xing, X., Babič, J., van Asseldonk, E. H. F., & Burdet, E. (2022) · Progress in Biomedical Engineering · doi:10.1088/2516-1091/ac8193 · Cited by 66
“While there have been great advances in perception and learning state representations in the field of deep learning that have enhanced RL [82], the expense and risk involved in pure exploration based data collection for real-world human–robot interactions is a significant bottleneck for deploying RL in real-world robo…”
BackgroundDeep Reinforcement Learning Verification: A Survey
Landers, M., & Doryab, A. (2023) · ACM Computing Surveys · doi:10.1145/3596444 · Cited by 79
“For example, a property might specify that for inputs �1 ∈ [2, 8] and �2 ∈ [6, 12], the output � ∈ [0, 100].”
Wang, X., Zhou, J., Qin, B., & Guo, L. (2023) · IEEE Transactions on Sustainable Energy · doi:10.1109/tste.2023.3287871 · Cited by 64
Reinforcement learning informs optimal treatment strategies to limit antibiotic resistance
Weaver, D. T., King, E. S., Maltas, J., & Scott, J. G. (2024) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2303165121 · Cited by 64
“Deep Q learning is a well-studied and characterized method of reinforcement learning, and is particularly suited to situations where very little a priori knowledge about the environment is available 34,47 .”
MethodLeveraging Deep Reinforcement Learning Technique for Intrusion Detection in SCADA Infrastructure
Mesadieu, F., Torre, D., & Chennamaneni, A. (2024) · IEEE Access · doi:10.1109/access.2024.3390722 · Cited by 71
Centralized and Distributed Deep Reinforcement Learning Methods for Downlink Sum-Rate Optimization
Khan, A. A., & Adve, R. (2020) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2020.3022705 · Cited by 70
Control Regularization for Reduced Variance Reinforcement LearningPreprint
Cheng, R., Verma, A., Orosz, G., Chaudhuri, S., Yue, Y., & Burdick, J. W. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.05380 · Cited by 96
“Several studies have noted this high variability in learning as a significant hurdle for the application of RL, since learning becomes unreliable (Henderson et al., 2018; Arulkumaran et al., 2017; Recht, 2019).”
BackgroundMeta-Learning Acquisition Functions for Transfer Learning in Bayesian OptimizationPreprint
Volpp, M., Fröhlich, L. P., Fischer, K., Doerr, A., Falkner, S., Hutter, F., & Daniel, C. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.02642 · Cited by 82
“ion and the random variables described by the surrogate model. (ˇ) E ˇ hP T t=0 tr t s 0 i . Here, Tdenotes the episode length and 2(0;1] is a discount factor. There exists a wide range of methods (Arulkumaran et al., 2017) for learning policies ˇthat maximize ( ). A recent and particularly successful algorithm is Pro…”
BackgroundA Comprehensive Survey of Machine Learning Applied to Radar Signal ProcessingPreprint
Lang, P., Fu, X., Martorella, M., Dong, J., Qin, R., Meng, X., & Xie, M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.13702 · Cited by 81
“Reinforcement learning (RL) system is also an unsupervised learning framework concerning on iteratively making optimal tactical actions act on environment to obtain maximum total amount of rewards [16].”
BackgroundInfluentialReinforcement Learning for Cyber-Physical Systems
Liu, X., Xu, H., Liao, W., & Yu, W. (2019) · IEEE International Conference on Industrial Internet (ICII) · doi:10.1109/icii.2019.00063 · Cited by 66
“Additionally, there are other surveys and tutorials about reinforcement learning methodologies and algorithms from different perspectives, including reinforcement learning in robotics [10], multiagent reinforcement learning, safe reinforcement learning, and deep reinforcement learning [16], [17].”
Yin, L., & Li, S. (2021) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2021.104326 · Cited by 63
Dangut, M. D., Jennions, I. K., King, S., & Skaf, Z. (2022) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2022.108873 · Cited by 62
“[23] Arulkumaran K, Deisenroth MP, Brundage M, Bharath AA.”
BackgroundMethodAnalog Integrated Circuit Topology Synthesis With Deep Reinforcement Learning
Zhao, Z., & Zhang, L. (2022) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2022.3153437 · Cited by 60
“There are several schemes that have been proposed to carry out the DRL process, such as Q -learning, policy gradient, and advantage actor–critic [31].”
BackgroundPeng, Z., Liu, E., Pan, C., Wang, H., Wang, D., & Liu, L. (2022) · Journal of the Franklin Institute · doi:10.1016/j.jfranklin.2022.10.020 · Cited by 61
Rokhforoz, P., Montazeri, M., & Fink, O. (2023) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2022.109081 · Cited by 60
Heo, J. C., & Chang, S. (2024) · Sustainable Cities and Society · doi:10.1016/j.scs.2024.105567 · Cited by 60
Wang, T., Ma, M., Liang, S., Yang, J., & Wang, Y. (2025) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2025.105005 · Cited by 67
Deep Reinforcement Learning: A State-of-the-Art Walkthrough
Lazaridis, A., Fachantidis, A., & Vlahavas, I. P. (2020) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.12412 · Cited by 78
“Similarly, Arulkumaran et al. (2017) developed a brief survey for Deep RL methods and algorithms that is also similar to our work, providing the basic theoretical background on RL formulation, introducing primary state-of-the-art Deep RL algorithms and discussing current research areas and…”
MethodGonzález-García, A., Barragan-Alcantar, D., Collado-Gonzalez, I., & Garrido, L. (2021) · Control Engineering Practice · doi:10.1016/j.conengprac.2021.104807 · Cited by 64
EconAgent: Large Language Model-Empowered Agents for Simulating Macroeconomic Activities
Li, N., Gao, C., Li, M., Li, Y., & Liao, Q. (2024) · Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2024.acl-long.829 · Cited by 59
“Besides, we select a learning-based method, AI-Economist (Zheng et al., 2022) AI-Eco , which builds on the assumption of rational decision-making and employs reinforcement learning (RL) (Arulkumaran et al., 2017) to maximize the agent’s utility.”
MethodDeep Reinforcement Learning for Trading—A Critical Survey
Millea, A. (2021) · Data · doi:10.3390/data6110119 · Cited by 74
Deep Adversarial Imitation Reinforcement Learning for QoS-Aware Cloud Job Scheduling
Huang, Y., Cheng, L., Xue, L., Liu, C., Li, Y., Li, J., & Ward, T. E. (2021) · IEEE Systems Journal · doi:10.1109/jsyst.2021.3122126 · Cited by 63
“In fact, without requiring any prior knowledge of a system, DRL has demonstrated its strong capability on making decisions for complex problems in many domains [35].”
BackgroundMulti objective dynamic task scheduling optimization algorithm based on deep reinforcement learning
Cheng, Y., Cao, Z., Zhang, X., Cao, Q., & Zhang, D. (2023) · The Journal of Supercomputing · doi:10.1007/s11227-023-05714-1 · Cited by 67
“Deep reinforcement learning not only uses deep learning to capture environmental features but also uses reinforcement learning to adapt to complex and changing dynamic scenarios [5], which further improves the ability of reinforcement learning to solve scheduling problems [14].”
Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
Nan, T., Zheng, S., Qiao, S., Quan, H., Gao, X., Niu, J., Zheng, B., Guo, C., Zhang, Y., Wang, X., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-60307-1 · Cited by 68
“Construction of an RL model to imitate the slide-reviewing behavior of the pathologists For a given WSI W , an RL 45,46 task can emulate the visual behavior of expert pathologists to conduct a rapid search on a two-dimensional plane and locate areas potentially harboring lesions 47,48 .”
Liu, Y., Zhang, W., Pan, S., Li, Y., & Chen, Y. (2019) · Computer Communications · doi:10.1016/j.comcom.2019.11.031 · Cited by 64
From Simulation to Reality: A Learning Framework for Fish-Like Robots to Perform Control Tasks
Zhang, T., Tian, R., Yang, H., Wang, C., Sun, J., Zhang, S., & Xie, G. (2022) · IEEE Transactions on Robotics · doi:10.1109/tro.2022.3181014 · Cited by 58
“two main categories [34]: methods based on value functions, such as the deep Q-network (DQN) [35], and methods based on policy search, such as DDPG [17].”
MethodIncentive learning-based energy management for hybrid energy storage system in electric vehicles
Li, F., Gao, Y., Wu, Y., Xia, Y., Wang, C., Hu, J., & Huang, Z. (2023) · Energy Conversion and Management · doi:10.1016/j.enconman.2023.117480 · Cited by 61
Deng, H., Zhao, Y., Wang, Q., & Nguyen, A. (2023) · Automotive Innovation · doi:10.1007/s42154-023-00231-6 · Cited by 61
“However, traditional reinforcement learning methods face challenges due to the high dimensionality of input features and computational complexity [43].”
BackgroundA Comprehensive Survey on Machine and Deep Learning for Optical Communications
Amirabadi, M. A., Nezamalhosseini, S. A., Kahaei, M. H., & Chen, L. R. (2025) · IEEE Access · doi:10.1109/access.2025.3569559 · Cited by 67
An, H., & Jung, J.-I. (2019) · Electronics · doi:10.3390/electronics8050543 · Cited by 61
Deep Reinforcement Learning Techniques in Diversified Domains: A Survey
Gupta, S., Singal, G., & Garg, D. (2021) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-021-09552-3 · Cited by 62
Ultrasound-Guided Robotic Navigation with Deep Reinforcement Learning
Hase, H., Azampour, M. F., Tirindelli, M., Paschali, M., Simson, W. A., Fatemizadeh, E., & Navab, N. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9340913 · Cited by 74
A bi-level cooperative operation approach for AGV based automated valet parking
Zhang, J., Li, Z., Li, L., Li, Y., & Dong, H. (2021) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2021.103140 · Cited by 60
Wireless Avionics Intracommunications: A Survey of Benefits, Challenges, and Solutions
Park, P., Marco, P. D., Nah, J., & Fischione, C. (2020) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.3038848 · Cited by 59
“Furthermore, reinforcement learning models are recently tied with many DL frameworks for resource allocation, routing, and network optimization since the communication system continuously interacts with the environment including the time-varying channel conditions [149], [147].”
BackgroundConsensus Algorithms and Deep Reinforcement Learning in Energy Market: A Review
Jogunola, O., Adebisi, B., Ikpehai, A., Popoola, S. I., Gui, G., Gacanin, H., & Ci, S. (2020) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.3032162 · Cited by 63
Deep reinforcement learning for conservation decisions
Lapeyrolerie, M., Chapman, M., Norman, K., & Boettiger, C. (2022) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.13954 · Cited by 62
Differential Privacy and IRS Empowered Intelligent Energy Harvesting for 6G Internet of Things
Pan, Q., Wu, J., Zheng, X., Yang, W., & Li, J. (2021) · IEEE Internet of Things Journal · doi:10.1109/jiot.2021.3104833 · Cited by 59
“As a mainstream of the AI technology, DRL has the strong capability to deal with the optimization problem in the complex and dynamic environment [36], [37].”
BackgroundArtificial Intelligence Application in Demand Response: Advantages, Issues, Status, and Challenges
Ali, A. N. F., Sulaima, M. F., Razak, I. A. W. A., Kadir, A. F. A., & Mokhlis, H. (2023) · IEEE Access · doi:10.1109/access.2023.3237737 · Cited by 64
“The RL framework has been utilized in a variety of fields, with robotics [74], resource management in computer clusters [75], playing video games via pixel input [76], and automated ML frameworks [77] being among the most significant.”
MethodHu, L., Han, C., Wang, X., Zhu, H., & Ouyang, J. (2024) · Sensors · doi:10.3390/s24061993 · Cited by 57
“These advancements have led to the emergence of deep reinforcement learning (DRL), a powerful tool that combines the representational ability of deep learning with the decision-making prowess of reinforcement learning [16].”
BackgroundPandey, V., Wang, E., & Boyles, S. D. (2020) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2020.102715 · Cited by 64
Lee, D., Arigi, A. M., & Kim, J. (2020) · IEEE Access · doi:10.1109/access.2020.3034218 · Cited by 57
Deep Reinforcement Learning Framework for Underwater Locomotion of Soft Robot
Li, G., Shintake, J., & Hayashibe, M. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561145 · Cited by 69
“As a solution to the problems mentioned above, deep reinforcement learning (DRL), which is a significant branch of machine learning, offers an end-to-end and model-free method that can solve the robot locomotion problem directly without the need for a complicated controller design [8]– [10].”
BackgroundWang, Z., Xiang, X., Duan, Y., & Yang, S. (2023) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2023.107728 · Cited by 59
Broad Reinforcement Learning for Supporting Fast Autonomous IoT
Wei, X., Zhao, J., Zhou, L., & Qian, Y. (2020) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.2980198 · Cited by 60
“[10].”
Tao, X., & Hafid, A. S. (2020) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.3022611 · Cited by 64
“In particular, they proposed a DRL-based scheme for network slicing.”
MethodInfluentialA Systematic Review on Reinforcement Learning-Based Robotics Within the Last Decade
Khan, M. A.-M., Khan, M. R. J., Tooshil, A., Sikder, N., Mahmud, M. A. P., Kouzani, A. Z., & Nahid, A.-A. (2020) · IEEE Access · doi:10.1109/access.2020.3027152 · Cited by 64
Kołodziejczyk, W., Żółtowska, I., & Cichosz, P. (2020) · Control Engineering Practice · doi:10.1016/j.conengprac.2020.104598 · Cited by 56
Yuan, T., Neto, W. B. D. R., Rothenberg, C. E., Obraczka, K., Barakat, C., & Turletti, T. (2020) · IEEE Transactions on Network and Service Management · doi:10.1109/tnsm.2020.3047765 · Cited by 58
“Deep Reinforcement Learning (DRL) [2], [19] is a machine learning technique that can efficiently tackle complex problems [20] such as dynamic resource allocation [21] by building a higher-level”
Introduction to Reinforcement LearningPreprint
Ghasemi, M., & Ebrahimi, D. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2408.07712 · Cited by 80
“As a result of this synergy, learning can be more stable and efficient, addressing some limitations of pure policy or Value-based approaches [16, 54].”
Svetozarevic, B., Baumann, C. A., Muntwiler, S., Natale, L. D., Zeilinger, M. N., & Heer, P. (2021) · Applied Energy · doi:10.1016/j.apenergy.2021.118127 · Cited by 59
Fang, Z., Jiang, D., Huang, J., Cheng, C., Sha, Q., He, B., & Li, G. (2022) · Ocean Engineering · doi:10.1016/j.oceaneng.2022.112182 · Cited by 55
Zhang, W., Zhao, T., Zhao, Z., Ma, D., & Liu, F. (2023) · Signal Processing · doi:10.1016/j.sigpro.2023.108965 · Cited by 55
A self-adaptive hardware with resistive switching synapses for experience-based neurocomputing
Bianchi, S., Muñoz-Martín, I., Covi, E., Bricalli, A., Piccolboni, G., Regev, A., Molas, G., Nodin, J. F., Andrieu, F., & Ielmini, D. (2023) · Nature Communications · doi:10.1038/s41467-023-37097-5 · Cited by 58
A Deep Reinforcement Learning-Based Preemptive Approach for Cost-Aware Cloud Job Scheduling
Cheng, L., Wang, Y., Cheng, F., Liu, C., Zhao, Z., & Wang, Y. (2023) · IEEE Transactions on Sustainable Computing · doi:10.1109/tsusc.2023.3303898 · Cited by 55
Zhou, Z., Wen, Z., Cui, J., Luan, H., Chen, Y., & Lin, D. (2025) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2025.110885 · Cited by 60
Reinforcement learning for vehicle-to-grid: A review
Xie, H., Song, G., Shi, Z., Zhang, J., Lin, Z., Yu, Q., Fu, H., Song, X., & Zhang, H. (2025) · Advances in Applied Energy · doi:10.1016/j.adapen.2025.100214 · Cited by 81
Crowd Navigation in an Unknown and Dynamic Environment Based on Deep Reinforcement Learning
Sun, L., Zhai, J., & Qin, W. (2019) · IEEE Access · doi:10.1109/access.2019.2933492 · Cited by 61
Energy optimization of electric vehicle’s acceleration process based on reinforcement learning
He, H., Cao, J., & Xing, C. (2019) · Journal of Cleaner Production · doi:10.1016/j.jclepro.2019.119302 · Cited by 57
Deep Q-Network based Anti-Jamming Strategy Design for Frequency Agile Radar
Li, K., Jiu, B., & Liu, H. (2019) · International Radar Conference (RADAR) · doi:10.1109/radar41533.2019.171227 · Cited by 54
“A MDP consists of a set of states S , a set of actions A , transition dynamics T ( s t +1 | s t , a t ) that describe how the state s t at time t transfers to s t +1 at time t +1 with action a t being taken and a reward function R ( s t , a t , s t +1 ) [12].”
BackgroundMethodQoS-Aware Task Offloading in Fog Environment Using Multi-agent Deep Reinforcement Learning
Jain, V., & Kumar, B. (2022) · Journal of Network and Systems Management · doi:10.1007/s10922-022-09696-y · Cited by 54
“An RL can be defined as Markov Decision Process which is represented by 5-tuple includes [45]:”
BackgroundApplying Artificial Intelligence in Cryptocurrency Markets: A Survey
Amirzadeh, R., Nazari, A., & Thiruvady, D. (2022) · Algorithms · doi:10.3390/a15110428 · Cited by 72
“One initial goal of AI is to create a fully autonomous agent capable of interacting with an environment for learning optimal behaviours that improve over time considering a particular objective [67].”
BackgroundA survey on blockchain sharding
Liu, X., Xie, H., Yan, Z., & Liang, X. (2023) · ISA Transactions · doi:10.1016/j.isatra.2023.06.029 · Cited by 56
Zhou, Z., Song, J., Xie, X., Shu, Z., Ma, L., Liu, D., Yin, J., & See, S. C. W. (2024) · IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP) · doi:10.1145/3639477.3639740 · Cited by 67
“To address these challenges, recent studies in robotics manipulation leverages AI techniques, e.g., DRL Arulkumaran et al. (2017), to learn a control policy directly from data, making it a typical and representative example of AI-CPS.”
MethodDeep Q-Network Learning Based Downlink Resource Allocation for Hybrid RF/VLC Systems
Shrivastava, S., Chen, B., Chen, C., Wang, H., & Dai, M. (2020) · IEEE Access · doi:10.1109/access.2020.3014427 · Cited by 59
Yao, J., Lu, B., & Zhang, J. (2021) · The International Journal of Advanced Manufacturing Technology · doi:10.1007/s00170-021-07950-2 · Cited by 57
A Review of Machine Learning Applications in FuzzingPreprint
Saavedra, G., Rodhouse, K., Dunlavy, D., & Kegelmeyer, P. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.11133 · Cited by 54
“[2] Kai Arulkumaran, Marc Peter Deisenroth, Miles Brundage, and”
Fang, H., Tu, Y., Wang, H., He, S., Liu, F., Ding, Z., & Cheng, S. S. (2022) · IEEE Transactions on Fuzzy Systems · doi:10.1109/tfuzz.2022.3171844 · Cited by 53
“INTRODUCTION IN REINFORCEMENT learning (RL), an agent tries to maximize the long-term reward by taking actions with respect to the internal state and external circumstances [1].”
BackgroundStiff-PDEs and Physics-Informed Neural Networks
Sharma, P., Evans, L. M., Tindall, M., & Nithiarasu, P. (2023) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-023-09890-4 · Cited by 110
“[70] proposed a sampling strategy for efficient training of PINNs based on an approximation method called importance sampling [80], which is often used in reinforcement learning for approximating the expected reward based on the older policy [81, 82].”
MethodParvaresh, N., & Kantarcı, B. (2023) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2023.3251297 · Cited by 61
“Since exploring an unlimited number of actions is not practical, the action space in QL and DQL should be limited to a finite number of actions, otherwise there exists the risk of converging to a local minimum by the agent [34].”
BackgroundNoack, B. R. (2023) · Physics of Fluids · doi:10.1063/5.0152777 · Cited by 53
A Remaining Useful Life Prediction Method of Rolling Bearings Based on Deep Reinforcement Learning
Zheng, G., Li, Y., Zhou, Z., & Yan, R. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3363610 · Cited by 55
“The agent learn to adjust its behavior based on received rewards in order to optimize for higher rewards [31].”
BackgroundZhang, Y., Mei, Z. X., Wu, X., Jiang, H., Zhang, J., & Gao, W. (2024) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2024.3399705 · Cited by 61
“While the above challenges could be well addressed with the development of reinforcement learning algorithm theory, and this type of approach has been used widely in many fields [17].”
BackgroundLi, H., Xiong, K., Lu, Y., Chen, W., Fan, P., & Letaief, K. B. (2024) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2024.3496536 · Cited by 57
Li, J., Liu, J., Yang, Q., Wang, T., He, H., Wang, H., & Sun, F. (2025) · Renewable and Sustainable Energy Reviews · doi:10.1016/j.rser.2025.115450 · Cited by 57
Unsupervised Representation Learning in Deep Reinforcement Learning: A Review
Botteghi, N., Poel, M., & Brüne, C. (2025) · IEEE Control Systems · doi:10.1109/mcs.2025.3534477 · Cited by 59
Wu, Q., Han, J., Yan, Y., Kuo, Y., & Shen, Z. M. (2025) · Health Care Management Science · doi:10.1007/s10729-025-09699-6 · Cited by 68
Multi-Agent Deep Reinforcement Learning for Multi-Object Tracker
Jiang, M., Hai, T., Pan, Z., Wang, H., Jia, Y., & Deng, C. (2019) · IEEE Access · doi:10.1109/access.2019.2901300 · Cited by 58
“Reinforcement learning (RL) has gained some success in the previous researches, but these existing approaches have poor scalability and are limited to low-dimensional issues [24]–[26].”
BackgroundReinforcement Learning Algorithms in Global Path Planning for Mobile Robot
Sichkar, V. N. (2019) · International Conference on Industrial Engineering, Applications and Manufacturing (ICIEAM) · doi:10.1109/icieam.2019.8742915 · Cited by 53
“In order to accomplish this goal Q-Learning algorithm uses Q-function, an argument of which is the action performed by the agent [1].”
MethodImproved day ahead heating demand forecasting by online correction methods
Bünning, F., Bollinger, A., Heer, P., Smith, R. S., & Lygeros, J. (2020) · Energy and Buildings · doi:10.1016/j.enbuild.2020.109821 · Cited by 56
Zhang, F., Han, G., Liu, L., Martínez-García, M., & Peng, Y. (2021) · IEEE Internet of Things Journal · doi:10.1109/jiot.2021.3068427 · Cited by 57
Path Following Control for UAV Using Deep Reinforcement Learning Approach
Zhang, Y., Zhang, Y. M., & Yu, Z. (2021) · Guidance Navigation and Control · doi:10.1142/s2737480721500059 · Cited by 56
Metadata
- Authors
- Arulkumaran, K., Deisenroth, M. P., Brundage, M., & Bharath, A. A.
- Affiliations
- University of Oxford (GB)
- Institution of Engineering and Technology (GB)
- Venue
- IEEE Signal Processing Magazine
- Published
- 2017-11
- Volume
- 34
- Issue
- 6
- Pages
- 26-38
- ISSN
- 1053-5888 (ISSN-L)
- OpenAlex
- W3100789280
- Cited by
- 4,733
- References
- 182
- Citations by year
- 2026: 444 · 2025: 855 · 2024: 806 · 2023: 749 · 2022: 590 · 2021: 511 · 2020: 388 · 2019: 260 · 2018: 122 · 2017: 6
- FWCI
- 126.58
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Signal Processing Magazine (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- UCL Discovery (University College London) (submittedVersion)
- Spiral (Imperial College London) (submittedVersion)
- arXiv (Cornell University)
- 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 reinforcement learning, value-based methods, deep Q-network, trust region policy optimization, asynchronous advantage actor-critic, visual understanding, autonomous systems
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.