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 901–1,000 of 1,021
Artificial intelligence for drug delivery: Yesterday, today and tomorrow
Wu, Y., Wang, N., Xiong, P., Wang, R., Deng, J., & Ouyang, D. (2025) · Acta Pharmaceutica Sinica B · doi:10.1016/j.apsb.2025.09.022 · Cited by 46
“Reinforcement learning (RL) is a machine learning paradigm specifically designed for optimization tasks 104 , aiming to learn a policy through interactions between an agent and the environment to maximize cumulative rewards.”
A Parallel Framework of Adaptive Dynamic Programming Algorithm With Off-Policy Learning
Sun, C., Li, X., & Sun, Y. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.3015767 · Cited by 39
5G and Beyond: Past, Present and Future of the Mobile Communications
Gutierrez, C. A., Rendón, O. M. C., & Campos-Delgado, D. U. (2021) · IEEE Latin America Transactions · doi:10.1109/tla.2021.9477273 · Cited by 43
Robustness and performance of Deep Reinforcement Learning
Al-Nima, R. R. O., Han, T., Al-Sumaidaee, S. A. M., Chen, T., & Woo, W. L. (2021) · Applied Soft Computing · doi:10.1016/j.asoc.2021.107295 · Cited by 44
“In other words, DL is structured within the RL to address Markov Decision Process (MDP) issues [2].”
Deep Reinforcement Learning for QoS provisioning at the MAC layer: A Survey
Abbasi, M., Shahraki, A., Piran, M. J., & Taherkordi, A. (2021) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2021.104234 · Cited by 44
Zhang, Y., & Zhao, Y. F. (2021) · Journal of Manufacturing Systems · doi:10.1016/j.jmsy.2021.07.002 · Cited by 42
Deep‐learning power and perspectives for genomic selection
Montesinos-López, O. A., Montesinos‐López, A., Hernández-Suárez, C., Barrón‐López, J. A., & Crossa, J. L. (2021) · The Plant Genome · doi:10.1002/tpg2.20122 · Cited by 39
“Nowadays, reinforcement learning has found application in problems from robotics to games and healthcare (Arulkumaran et al., 2017; Mnih et al., 2015).”
BackgroundApplications of Multi-Agent Deep Reinforcement Learning: Models and Algorithms
Ibrahim, A. M., Yau, K. A., Chong, Y.-W., & Wu, C. (2021) · Applied Sciences · doi:10.3390/app112210870 · Cited by 44
Asghari, A., & Sohrabi, M. K. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3161950 · Cited by 41
“Since the CBS placement in each zone is independent, deep Q -network (DQN) [18] and Markov game (MG) [19] are utilized to obtain a global pattern of RP. Finally, in the area covered by each CBS, more than one server is allocated, shared between adjacent areas and increases the reliability and the…”
MethodYu, P., Ding, Y., Li, Z., Tian, J., Zhang, J., Liu, Y., Li, W., & Qiu, X. (2022) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2022.3198834 · Cited by 41
Deep RL at Scale: Sorting Waste in Office Buildings with a Fleet of Mobile Manipulators
Herzog, A., Rao, K., Hausman, K., Lü, Y., Wohlhart, P., Yan, M., Lin, J., Arenas, M. G., Xiao, T. G., Kappler, D., et al. (2023) · Robotics: Science and Systems · doi:10.15607/rss.2023.xix.022 · Cited by 49
Optimization Methods for Interpretable Differentiable Decision Trees in Reinforcement LearningPreprint
Silva, A., Killian, T. W., Rodriguez, I. D. J., Son, S.-H., & Gombolay, M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.09338 · Cited by 43
A Survey on Quantum Reinforcement Learning
Meyer, N., Ufrecht, C., Periyasamy, M., Scherer, D. D., Plinge, A., & Mutschler, C. (2022) · arXiv · doi:10.48550/arxiv.2211.03464 · Cited by 38
Musaddiq, A., Olsson, T., & Ahlgren, F. (2023) · Sensors · doi:10.3390/s23198263 · Cited by 47
“In addition, the integration of deep learning with RL (DRL) has demonstrated an improved performance in complex control tasks [32].”
BackgroundLi, J., Shi, H., & Hwang, K. (2023) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2023.3323307 · Cited by 40
“Therefore, the literature of robotics aimed at model-free reinforcement learning technologies [40].”
BackgroundGajcin, J., & Dusparić, I. (2024) · ACM Computing Surveys · doi:10.1145/3648472 · Cited by 51
“Lack of transparency of AI models can cause them to inadvertently adopt historical biases in-grained in training data and use them in their decision logic [39].”
BackgroundDavid, H. A., Bahrpeyma, F., & Reichelt, D. (2024) · Journal of Manufacturing Systems · doi:10.1016/j.jmsy.2024.09.019 · Cited by 38
Load frequency control based on reinforcement learning for microgrids under false data attacks
Abouzeid, S. I., Chen, Y., Zaery, M., Abido, M. A., Raza, A., & Abdelhameed, E. H. (2025) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2025.110093 · Cited by 38
Trading-Off Accuracy and Energy of Deep Inference on Embedded Systems: A Co-Design Approach
Jayakodi, N. K., Chatterjee, A., Choi, W., Doppa, J. R., & Pande, P. P. (2018) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2018.2857338 · Cited by 39
Hasan, M. M., Lwin, K. T., Imani, M., Shabut, A. M., Bittencourt, L. F., & Hossain, M. A. (2019) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2019.08.014 · Cited by 41
“On the other hand, there has been a growing interest (Arulkumaran et al., 2017) to solve the multi-objective optimisation in sequential decision making using reinforcement learning especially deep reinforcement learning after the success of DeepMind in 2015 (Mnih et al., 2015).”
BackgroundDeep Learning in Aquaculture: A Review
Sun, M., Yang, X., & Xie, Y. (2020) · Cited by 71
Improve PID controller through reinforcement learning
Qin, Y., Zhang, W., Shi, J., & Liu, J. (2018) · IEEE CSAA Guidance, Navigation and Control Conference (CGNCC) · doi:10.1109/gncc42960.2018.9019095 · Cited by 46
Hwangbo, S., & Sin, G. (2020) · Computers & Chemical Engineering · doi:10.1016/j.compchemeng.2020.106910 · Cited by 38
Schilling, M., Konen, K., Ohl, F. W., & Korthals, T. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341754 · Cited by 44
Zhou, P., Xie, Y., Niu, B., Pu, L., Xu, Z., Jiang, H., & Huang, H. (2020) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2020.3038998 · Cited by 39
“By further embracing the any nonlinear expression capability of deep neural networks, we then find that Deep Reinforcement Learning (DRL) can solve decision problems that are difficult for RL to solve, i.e., the problem of high-dimensional state space and action space settings [10].”
BackgroundImpact of the reperfusion status for predicting the final stroke infarct using deep learning
Debs, N., Cho, T., Rousseau, D., Berthezène, Y., Buisson, M., Eker, O., Mechtouff, L., Nighoghossian, N., Ovize, M., & Frindel, C. (2020) · NeuroImage Clinical · doi:10.1016/j.nicl.2020.102548 · Cited by 42
“Reinforcement learning could help improve the performance of CNNs by training more specifically on these underrepresented patients (Arulkumaran et al., 2017).”
BackgroundHypernetwork Dismantling via Deep Reinforcement Learning
Yan, D., Xie, W., Zhang, Y., He, Q., & Yang, Y. J. (2022) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2022.3174163 · Cited by 40
“Deep reinforcement learning [27], [28] is a promising approach to solve the high-dimension issue in reinforcement learning through the powerful representation ability of deep learning.”
BackgroundToward Multi-FPGA Acceleration of the Neural Networks
Biookaghazadeh, S., Ravi, P. K., & Zhao, M. (2021) · ACM Journal on Emerging Technologies in Computing Systems · doi:10.1145/3432816 · Cited by 40
“CNNs are the main building blocks in many AI applications, such as image classification [7], reinforcement learning [8], and natural language processing.”
BackgroundMobility Load Management in Cellular Networks: A Deep Reinforcement Learning Approach
Alsuhli, G. H., Banawan, K., Attiah, K. M., Elezabi, A. Y., Seddik, K. G., Gaber, A., Zaki, M., & Gadallah, Y. (2021) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2021.3107458 · Cited by 46
How to compete with robots by assessing job automation risks and resilient alternatives
Paolillo, A., Colella, F., Nosengo, N., Schiano, F., Stewart, W., Zambrano, D., Chappuis, I., Lalive, R., & Floreano, D. (2022) · Science Robotics · doi:10.1126/scirobotics.abg5561 · Cited by 49
Accelerated design for magnetic high entropy alloys using data-driven multi-objective optimization
Li, X., Shan, G., Zhang, J., & Shek, C. H. (2022) · Journal of Materials Chemistry C · doi:10.1039/d2tc03922b · Cited by 40
Chen, X., Dai, W., Ni, W., Wang, X., Zhang, S., Xu, S., & Sun, Y. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3251353 · Cited by 38
“The model can better converge by exploring different environment states [34].”
BackgroundArishi, A., & Krishnan, K. (2023) · Journal of Management Analytics · doi:10.1080/23270012.2023.2229842 · Cited by 56
Alonso, R. S., Sittón-Candanedo, I., Vara, R. C., Prieto, J., & Corchado, J. M. (2020) · Sustainability · doi:10.3390/su12145706 · Cited by 40
Yang, F., Yang, Y., Ni, S., Liu, S., Xu, C., Chen, D., & Zhang, Q. (2023) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2023.104237 · Cited by 37
Bouchmal, O., Cimoli, B., Stabile, R., Olmos, J. J. V., & Monroy, I. T. (2023) · Frontiers in Communications and Networks · doi:10.3389/frcmn.2023.1220227 · Cited by 41
Klein, L., Zelinka, I., & Seidl, D. (2024) · Swarm and Evolutionary Computation · doi:10.1016/j.swevo.2024.101487 · Cited by 40
Li, Y., He, H., Khajepour, A., Chen, Y., Huo, W., & Wang, H. (2024) · IEEE Transactions on Transportation Electrification · doi:10.1109/tte.2024.3377809 · Cited by 46
Mahjoub, C., Hamdi, M., Alkanhel, R. I., Mohamed, S., & Ejbali, R. (2024) · EURASIP Journal on Wireless Communications and Networking · doi:10.1186/s13638-024-02348-6 · Cited by 44
“[8].”
BackgroundWang, C., Wang, J., Ma, Z., Xu, M., Qi, K., Ji, Z., & Wei, C. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3425755 · Cited by 40
A Futuristic Development in 3D Printing Technique Using Nanomaterials with a Step Toward 4D Printing
Agarwal, P., Mathur, V., Kasturi, M., Srinivasan, V., Seetharam, R. N., & Vasanthan, K. S. (2024) · ACS Omega · doi:10.1021/acsomega.4c04123 · Cited by 40
Dual Network Computation Offloading Based on DRL for Satellite-Terrestrial Integrated Networks
Li, D., Sun, Y., Peng, J., Cheng, S., Yin, Z., Cheng, N., Liu, J., Li, Z., & Xu, C. (2024) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2024.3493388 · Cited by 39
A First Look at Generative Artificial Intelligence-Based Music Therapy for Mental Disorders
Shen, L., Zhang, H., Zhu, C., Li, R., Qian, K., Meng, W., Tian, F., Hu, B., Schuller, B. W., & Yamamoto, Y. (2024) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2024.3514633 · Cited by 44
Advancements in ultrafast photonics: confluence of nonlinear optics and intelligent strategies
Wu, Q. Y., Peng, L., Huang, Z., Liu, X., Luo, M., Gao, D., & Meng, H. (2025) · Light Science & Applications · doi:10.1038/s41377-024-01732-7 · Cited by 48
Continual Deep Reinforcement Learning for Decentralized Satellite Routing
Lozano-Cuadra, F., Soret, B., Leyva‐Mayorga, I., & Popovski, P. (2025) · IEEE Transactions on Communications · doi:10.1109/tcomm.2025.3562522 · Cited by 46
“As the complexity of the environment increases, DRL is a particular type of RL that exploits the power of multi layer deep neural networks for state representation and/or function approximations [6].”
FedAgent: Federated Learning on Non-IID Data via Reinforcement Learning and Knowledge Distillation
Sun, B., Song, X., Tu, Y., & Liu, M. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.127973 · Cited by 43
Collision-avoidance under COLREGS for unmanned surface vehicles via deep reinforcement learning
Ma, Y., Zhao, Y., Wang, Y., Gan, L., & Zheng, Y. (2020) · Maritime Policy & Management · doi:10.1080/03088839.2020.1756494 · Cited by 37
“Referring to DRL algorithm, it is the combination of DL algorithm and RL algorithm (Arulkumaran et al. 2017), and is featured by autonomous interpretation and learning ability of RL algorithm, and is fused with environmental state perception and prediction ability of neural network (Zhao et al.”
MethodRecurrent MADDPG for Object Detection and Assignment in Combat Tasks
Wei, X., Yang, L., Cao, G., Lü, T., & Wang, B. (2020) · IEEE Access · doi:10.1109/access.2020.3022638 · Cited by 40
Towards closing the sim-to-real gap in collaborative multi-robot deep reinforcement learning
Zhao, W., Queralta, J. P., Qingqing, L., & Westerlund, T. (2020) · 2020 5th International Conference on Robotics and Automation Engineering (ICRAE) · doi:10.1109/icrae50850.2020.9310796 · Cited by 36
“Reinforcement learning (RL) algorithms for robotics and cyber-physical systems have seen an increasing adoption across multiple domains over the past decade [1], [2].”
Adaptive deep reinforcement learning for non-stationary environments
Zhu, J. S., Wei, Y., Kang, Y., Jiang, X., & Dullerud, G. E. (2022) · Science China Information Sciences · doi:10.1007/s11432-021-3347-8 · Cited by 40
“Among the existing DRL algorithms, the deep Q -learning algorithm is a classic and effective one, especially for MDP with large state spaces [33].”
MethodPopescu, D., Ichim, L., & Stoican, F. (2023) · Frontiers in Plant Science · doi:10.3389/fpls.2023.1237695 · Cited by 38
“More recently, they can also be learned effectively through the reinforcement learning method (Arulkumaran et al., 2017) and federated learning (Deng et al., 2022).”
Wu, S., Tian, D., Duan, X., Zhou, J., Zhao, D., & Cao, D. (2024) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2024.3380074 · Cited by 39
Integrated reinforcement and imitation learning for tower crane lift path planning
Wang, Z., Huang, C. Y., Yao, B., & Li, X. (2024) · Automation in Construction · doi:10.1016/j.autcon.2024.105568 · Cited by 41
Das, B. R., Hasan, S. R., Sabuj, S. R., Hossain, M. A., & Ray, S. K. (2025) · International Journal of Intelligent Networks · doi:10.1016/j.ijin.2025.06.001 · Cited by 49
A Q-Learning-Based Approach for Deploying Dynamic Service Function Chains
Sun, J., Huang, G., Sun, G., Yu, H., Sangaiah, A. K., & Chang, V. I. C. (2018) · Symmetry · doi:10.3390/sym10110646 · Cited by 36
“However, there are many researches related to reinforcement learning, such as stochastic learning automata [21,22], Q-learning [23], deep Q network [24], and Generative Adversarial Networks [25].”
BackgroundOn the Crossroad of Artificial Intelligence: A Revisit to Alan Turing and Norbert Wiener
Li, L., Zheng, N., & Wang, F. (2018) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2018.2884315 · Cited by 40
“reinforcement learning has attracted increasing attention [37].”
Continuous drone control using deep reinforcement learning for frontal view person shooting
Passalis, N., & Tefas, A. (2019) · Neural Computing and Applications · doi:10.1007/s00521-019-04330-6 · Cited by 39
Deep Reinforcement Learning for Network Selection Over Heterogeneous Health Systems
Chkirbene, Z., Abdellatif, A. A., Mohamed, A. M. S., Erbad, A., & Guizani, M. (2021) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2021.3058037 · Cited by 40
META: A City-Wide Taxi Repositioning Framework Based on Multi-Agent Reinforcement Learning
Liu, C., Chen, C., & Chen, C. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3096226 · Cited by 36
Achieving Zero Constraint Violation for Constrained Reinforcement Learning via Primal-Dual Approach
Bai, Q., Bedi, A. S., Agarwal, M., Koppel, A., & Aggarwal, V. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i4.20281 · Cited by 86
Dong, P., Chen, Z., Liao, X., & Yu, W. (2021) · Petroleum Science · doi:10.1016/j.petsci.2021.09.046 · Cited by 36
Reinforcement learning for resource management in multi-tenant serverless platforms
Qiu, H., Mao, W., Patke, A., Wang, C., Franke, H., Kalbarczyk, Z., Başar, T., & Iyer, R. K. (2022) · Workshop on Machine Learning and Systems (EuroMLSys) · doi:10.1145/3517207.3526971 · Cited by 36
Deep Learning, Blockchain based Multi-layered Authentication and Security Architectures
Sheth, H. S. K., Ilavarasi, A. K., & Tyagi, A. K. (2022) · International Conference on Applied Artificial Intelligence and Computing (ICAAIC) · doi:10.1109/icaaic53929.2022.9793179 · Cited by 40
“IDSs are also utilized to recognize a specific type of attack, such as a distributed denial-of-service (DDoS) attack.[10] Artificial Intelligence-based systems, on the other hand, have their own set of problems; false positives can be far more common than genuine threats.”
BackgroundTransfer Reinforcement Learning for Adaptive Task Offloading Over Distributed Edge Clouds
Shuai, K., Miao, Y., Hwang, K., & Li, Z. (2022) · IEEE Transactions on Cloud Computing · doi:10.1109/tcc.2022.3192560 · Cited by 37
“Asynchronous advantage actor critic . devices collectively train a single reinforcement learning model (an actor and a critic) for offloading. dA2C 1 [36].”
BackgroundMethodInfluentialReinforcement Learning: Theory and Applications in HEMS
Al-Ani, O., & Das, S. (2022) · Energies · doi:10.3390/en15176392 · Cited by 41
Automation of membrane capacitive deionization process using reinforcement learning
Yoon, N., Park, S., Son, M., & Cho, K. H. (2022) · Water Research · doi:10.1016/j.watres.2022.119337 · Cited by 37
Pereira, H., Gomes, L., & Vale, Z. (2022) · Energy Informatics · doi:10.1186/s42162-022-00235-2 · Cited by 36
“DRL handles standard RL challenges by performing complex tasks with less prior knowledge, as a result of its ability to learn abstraction levels from data (Arulkumaran et al. 2017).”
BackgroundMirageNet - Towards a GAN-based Framework for Synthetic Network Traffic Generation
Nukavarapu, S. K., Ayyat, M., & Nadeem, T. (2022) · Globecom · doi:10.1109/globecom48099.2022.10001494 · Cited by 36
“Similarly, a post-processing tool might be created by employing a reinforcement learning (RL) agent [3] capable”
MethodA predictive target tracking framework for IoT using CNN–LSTM
Hussain, L. A., Singh, S., Mizouni, R., Otrok, H., & Damiani, E. (2023) · Internet of Things · doi:10.1016/j.iot.2023.100744 · Cited by 36
Optimal active particle navigation meets machine learning (a)
Nasiri, M., Löwen, H., & Liebchen, B. (2023) · Europhysics Letters (EPL) · doi:10.1209/0295-5075/acc270 · Cited by 42
Unbalanced Web Phishing Classification through Deep Reinforcement Learning
Maci, A., Santorsola, A., Coscia, A., & Iannacone, A. (2023) · Computers · doi:10.3390/computers12060118 · Cited by 39
“Therefore, Q-learning becomes unreliable in practice [4].”
MethodMachine Learning and Quantum DevicesPreprint
Marquardt, F. (2021) · MPG.PuRe (Max Planck Society) · Cited by 35
Deep Reinforcement Learning Based Traffic Signal Control: A Comparative Analysis
Wu, C., Kim, I., & Ma, Z. (2023) · Procedia Computer Science · doi:10.1016/j.procs.2023.03.036 · Cited by 40
Sharma, P., Gupta, S., Bhatia, V., & Prakash, S. (2023) · IET Quantum Communication · doi:10.1049/qtc2.12063 · Cited by 38
A systematic review of federated learning incentive mechanisms and associated security challenges
Ali, A., Ilahi, I., Qayyum, A., Mohammed, I., Al‐Fuqaha, A., & Qadir, J. (2023) · Computer Science Review · doi:10.1016/j.cosrev.2023.100593 · Cited by 43
A review of automation of laser optics alignment with a focus on machine learning applications
Rakhmatulin, I., Risbridger, D. R., Carter, R., Esser, M. J. D., & Erden, M. S. (2023) · Optics and Lasers in Engineering · doi:10.1016/j.optlaseng.2023.107923 · Cited by 35
“In recent years, there has been a breakthrough in the field of reinforcement learning [80, 81], primarily due to earlier success when applied to various games [82], with it now being widely used to control robots [83,84].”
BackgroundA review of dialogue systems: current trends and future directions
Algherairy, A. S., & Ahmed, M. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-09322-1 · Cited by 59
“The idea is to parameterized the policy itself, and updated those parameters to maximize the expected return [78].”
InfluentialImproved Exploration Strategy for Q-Learning Based Multipath Routing in SDN Networks
Hassen, H., Meherzi, S., & Jemaa, Z. B. (2024) · Journal of Network and Systems Management · doi:10.1007/s10922-024-09804-0 · Cited by 35
Reinforcement-Learning-Based Path Planning: A Reward Function Strategy
Jaramillo-Martínez, R., Chavero-Navarrete, E., & Ibarra-Pérez, T. (2024) · Applied Sciences · doi:10.3390/app14177654 · Cited by 35
“The number of turns for the ID3QN and DQN algorithms and the proposed DQL method is 7, 4, and 4, respectively.”
Zhao, S., Gong, S., Gu, B., Li, L., Lyu, B., Hoang, D. T., & Yi, C. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3522249 · Cited by 40
“Hence, we update the actor network’s parameter ϕ i by using the deterministic policy gradient method [35], and the derivative of J i ( ϕ i ) is represented as follows : The i -th critic network is updated by the temporal difference (TD) error.”
Rezaei, M., & Nezamabadi–pour, H. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-024-11066-w · Cited by 42
Ji, E., Wang, Y., Xing, S., & Jin, J. (2025) · IEEE Access · doi:10.1109/access.2025.3598712 · Cited by 94
“This configuration was determined through systematic evaluation of attention mechanisms with varying head counts (2, 4, 8, 16), where 4 heads provided optimal balance between representational capacity and computational efficiency for capturing cross-datacenter relationships.”
Toward Autonomous UAV Swarm Navigation: A Review of Trajectory Design Paradigms
Arshid, K., Krayani, A., Marcenaro, L., Martín, D., & Regazzoni, C. S. (2025) · Sensors · doi:10.3390/s25185877 · Cited by 43
“DRL is a modern learning method where an agent observes the environment, performs actions, and improves its policy based on rewards [20,125].”
Semantic Analysis for Deep Q-Network in Android GUI Testing
Vuong, T. A. T., & Takada, S. (2019) · Proceedings/Proceedings of the ... International Conference on Software Engineering and Knowledge Engineering · doi:10.18293/seke2019-080 · Cited by 35
Applying Deep-Q Networks to Target Tracking to Improve Cognitive Radar
Kozy, M. A., Yu, J., Buehrer, R. M., Martone, A. F., & Sherbondy, K. D. (2019) · IEEE Radar Conference (RadarConf) · doi:10.1109/radar.2019.8835780 · Cited by 35
Zhao, J., Zhang, Y., Nie, Y., & Liu, J. (2020) · IEEE Access · doi:10.1109/access.2019.2963751 · Cited by 35
Application of deep neural network and deep reinforcement learning in wireless communication
Li, M., & Li, H. (2020) · PLoS ONE · doi:10.1371/journal.pone.0235447 · Cited by 35
RetractedPartially Explainable Big Data Driven Deep Reinforcement Learning for Green 5G UAV
Guo, W. (2020) · International Conference on Communications · doi:10.1109/icc40277.2020.9149151 · Cited by 38
“In DRL (see Fig 2a), the key aspects are: (i) state, (ii) action, (iii) reward, (iv) a model of environment dynamics, (v) policy and (vii) DNN implementation [14].”
BackgroundA Coach-Based Bayesian Reinforcement Learning Method for Snake Robot Control
Jia, Y., & Ma, S. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3061372 · Cited by 35
“Simulation results using the proposed CBRL with comparisons of MFRL [16] and state-of-the-art DLDC [10].”
MethodInfluentialUSV Path-Following Control Based On Deep Reinforcement Learning and Adaptive Control
González-García, A., Castañeda, H., & Garrido, L. (2020) · Global Oceans 2020: Singapore – U.S. Gulf Coast · doi:10.1109/ieeeconf38699.2020.9389360 · Cited by 37
“These functions express the expected discounted reward after taking an action a t in state s t , and follows the policy π , as: The combination of RL with DL establishes the field of DRL [28].”
BackgroundSelf-optimizing adaptive optics control with reinforcement learning for high-contrast imaging
Landman, R., Haffert, S. Y., Radhakrishnan, V. M., & Keller, C. U. (2021) · Journal of Astronomical Telescopes Instruments and Systems · doi:10.1117/1.jatis.7.3.039002 · Cited by 35
Cheng, M., & Frangopol, D. M. (2021) · Journal of Computing in Civil Engineering · doi:10.1061/(asce)cp.1943-5487.0000991 · Cited by 35
Zhang, X., & Wang, Y. (2022) · Applied Intelligence · doi:10.1007/s10489-022-03482-8 · Cited by 36
Chen, W. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.04.131 · Cited by 39
Learning to Selectively Transfer: Reinforced Transfer Learning for Deep Text Matching
Qu, C., Ji, F., Qiu, M., Yang, L., Min, Z., Chen, H., Huang, J., & Croft, W. B. (2019) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3289600.3290978 · Cited by 35
“mmerce Domain) I need to understand why my orders have been cancelled Why my order have been closed? ReinforcementLearning.The concept of reinforcement learning (RL) dates back to several decades ago [1, 31]. Recent advances in deep learning made it possible for RL agents to generate dialogs [16], play video games [18…”
BackgroundNew challenges in reinforcement learning: a survey of security and privacy
Lei, Y., Ye, D., Shen, S., Sui, Y., Zhu, T., & Zhou, W. (2022) · Artificial Intelligence Review · doi:10.1007/s10462-022-10348-5 · Cited by 39
“It can ensure that any individual record being stored in or removed from a dataset makes little difference to the dataset’s analytical output (Arulkumaran et al. 2017). The work of Wang and Hegde (2019) considered the use of differential private methods for reinforcement learning in continuous spaces, with a focus on…”
BackgroundInfluentialWhen machine learning meets Network Management and Orchestration in Edge-based networking paradigms
Shahraki, A., Ohlenforst, T., & Kreyß, F. (2022) · Journal of Network and Computer Applications · doi:10.1016/j.jnca.2022.103558 · Cited by 38
Zhang, K., Lin, X., & Li, M. (2023) · Physica A Statistical Mechanics and its Applications · doi:10.1016/j.physa.2023.128451 · Cited by 35
Lobbezoo, A., & Kwon, H. (2023) · Robotics · doi:10.3390/robotics12010012 · Cited by 35
“Compared to other RL applications such as self-driving cars and video games [14–16], robotic control is difficult due to the high dimensionality of the problem and the continuous space of actions [17–19].”
Learning Dexterous Manipulation from Exemplar Object Trajectories and Pre-Grasps
Dasari, S., Gupta, A., & Kumar, V. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161147 · Cited by 63
“As a result, the field has trended towards robot learning paradigms [27, 4, 17], which seek to leverage data-driven exploration to automatically acquire dexterous behaviors [25, 24, 39].”
Background
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.