{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T00:57:03Z","timestamp":1772067423961,"version":"3.50.1"},"reference-count":67,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"9","license":[{"start":{"date-parts":[[2025,9,1]],"date-time":"2025-09-01T00:00:00Z","timestamp":1756684800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2025,9,1]],"date-time":"2025-09-01T00:00:00Z","timestamp":1756684800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,9,1]],"date-time":"2025-09-01T00:00:00Z","timestamp":1756684800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62125305"],"award-info":[{"award-number":["62125305"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["U23A20339"],"award-info":[{"award-number":["U23A20339"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62088102"],"award-info":[{"award-number":["62088102"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62203348"],"award-info":[{"award-number":["62203348"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["52435010"],"award-info":[{"award-number":["52435010"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100007128","name":"Natural Science Foundation of Shaanxi Province","doi-asserted-by":"publisher","award":["2025SYS-SYSZD-083"],"award-info":[{"award-number":["2025SYS-SYSZD-083"]}],"id":[{"id":"10.13039\/501100007128","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Syst. Man Cybern, Syst."],"published-print":{"date-parts":[[2025,9]]},"DOI":"10.1109\/tsmc.2025.3578050","type":"journal-article","created":{"date-parts":[[2025,6,27]],"date-time":"2025-06-27T13:46:31Z","timestamp":1751031991000},"page":"6164-6176","source":"Crossref","is-referenced-by-count":2,"title":["Improving Sample Efficiency Through Stability Enhancement in Deep-Reinforcement Learning"],"prefix":"10.1109","volume":"55","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6068-6978","authenticated-orcid":false,"given":"Ziru","family":"Wang","sequence":"first","affiliation":[{"name":"College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-7121-828X","authenticated-orcid":false,"given":"Wanli","family":"Jiang","sequence":"additional","affiliation":[{"name":"College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an, China"}]},{"given":"Ru","family":"Peng","sequence":"additional","affiliation":[{"name":"College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-5797-8915","authenticated-orcid":false,"given":"Qian","family":"Kou","sequence":"additional","affiliation":[{"name":"College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-5679-056X","authenticated-orcid":false,"given":"Lipeng","family":"Wan","sequence":"additional","affiliation":[{"name":"College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3422-944X","authenticated-orcid":false,"given":"Xuguang","family":"Lan","sequence":"additional","affiliation":[{"name":"College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an, China"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1038\/nature24270"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1109\/TSMC.2021.3098451"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/TSMC.2022.3209070"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/TSMC.2018.2870983"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/TSMC.2023.3346401"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/TSMC.2022.3230666"},{"key":"ref7","article-title":"DeepMind control suite","author":"Tassa","year":"2018","journal-title":"arXiv:1801.00690"},{"key":"ref8","first-page":"1861","article-title":"Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Haarnoja"},{"key":"ref9","first-page":"1","article-title":"Image augmentation is all you need: Regularizing deep reinforcement learning from pixels","volume-title":"Proc. 9th Int. Conf. Learn. Represent.","author":"Yarats"},{"key":"ref10","first-page":"1","article-title":"Mastering visual continuous control: Improved data-augmented reinforcement learning","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Yarats"},{"key":"ref11","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Brown"},{"issue":"240","key":"ref12","first-page":"1","article-title":"PaLM: Scaling language modeling with pathways","volume":"24","author":"Chowdhery","year":"2023","journal-title":"J. Mach. Learn. Res."},{"key":"ref13","article-title":"Prioritized experience replay","author":"Schaul","year":"2015","journal-title":"arXiv:1511.05952"},{"key":"ref14","first-page":"11790","article-title":"Large batch experience replay","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Lahire"},{"key":"ref15","first-page":"1","article-title":"Prioritizing samples in reinforcement learning with reducible loss","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"36","author":"Sujit"},{"key":"ref16","article-title":"Prioritized sequence experience replay","author":"Brittain","year":"2019","journal-title":"arXiv:1905.12726"},{"key":"ref17","article-title":"Topological experience replay","author":"Hong","year":"2022","journal-title":"arXiv:2203.15845"},{"key":"ref18","first-page":"15630","article-title":"Prioritized training on points that are learnable, worth learning, and not yet learnt","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Mindermann"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1002\/eng2.12298"},{"key":"ref20","first-page":"36026","article-title":"Trap and replace: Defending backdoor attacks by trapping them into an easy-to-replace subnetwork","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"35","author":"Wang"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1016\/j.ins.2022.11.148"},{"key":"ref22","first-page":"499","article-title":"Stability and generalization","volume":"2","author":"Bousquet","year":"2002","journal-title":"J. Mach. Learn. Res."},{"key":"ref23","first-page":"1225","article-title":"Train faster, generalize better: Stability of stochastic gradient descent","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Hardt"},{"key":"ref24","first-page":"3174","article-title":"Train simultaneously, generalize better: Stability of gradient-based minimax learners","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Farnia"},{"key":"ref25","first-page":"5809","article-title":"Fine-grained analysis of stability and generalization for stochastic gradient descent","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Lei"},{"key":"ref26","first-page":"6175","article-title":"Stability and generalization of stochastic gradient methods for minimax problems","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Lei"},{"key":"ref27","first-page":"745","article-title":"Stability and generalization of learning algorithms that converge to global optima","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Charles"},{"key":"ref28","first-page":"37735","article-title":"Stability and generalization for Markov chain stochastic gradient methods","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"35","author":"Wang"},{"key":"ref29","first-page":"14219","article-title":"An equivalence between loss functions and non-uniform sampling in experience replay","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Fujimoto"},{"key":"ref30","article-title":"Proximal policy optimization algorithms","author":"Schulman","year":"2017","journal-title":"arXiv:1707.06347"},{"key":"ref31","first-page":"2048","article-title":"Leveraging procedural generation to benchmark reinforcement learning","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Cobbe"},{"key":"ref32","volume-title":"Markov Decision Processes: Discrete Stochastic Dynamic Programming","author":"Puterman","year":"2014"},{"key":"ref33","first-page":"1","article-title":"Continuous control with deep reinforcement learning","volume-title":"Proc. 4th Int. Conf. Learn. Represent.","author":"Lillicrap"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1038\/nature14236"},{"key":"ref35","first-page":"1","article-title":"Dream to control: Learning behaviors by latent imagination","volume-title":"Proc. 8th Int. Conf. Learn. Represent.","author":"Hafner"},{"key":"ref36","first-page":"1","article-title":"Mastering Atari with discrete world models","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Hafner"},{"key":"ref37","article-title":"Model-invariant state abstractions for model-based reinforcement learning","author":"Tomar","year":"2021","journal-title":"arXiv:2102.09850"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v35i12.17276"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.2307\/j.ctt4cgngj.10"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/UR61395.2024.10597462"},{"key":"ref41","first-page":"1162","article-title":"Abstract value iteration for hierarchical reinforcement learning","volume-title":"Proc. Int. Conf. Artif. Intell. Stat.","author":"Jothimurugan"},{"key":"ref42","first-page":"1","article-title":"Sample complexity of goal-conditioned hierarchical reinforcement learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"36","author":"Robert"},{"issue":"330","key":"ref43","first-page":"1","article-title":"Goal-space planning with subgoal models","volume":"25","author":"Lo","year":"2024","journal-title":"J. Mach. Learn. Res."},{"key":"ref44","first-page":"10070","article-title":"Striving for simplicity and performance in off-policy DRL: Output normalization and non-uniform sampling","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Wang"},{"key":"ref45","first-page":"110","article-title":"Experience replay with likelihood-free importance weights","volume-title":"Proc. Learn. Dyn. Control Conf.","author":"Sinha"},{"key":"ref46","first-page":"18560","article-title":"DisCor: Corrective feedback in reinforcement learning via distribution correction","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Kumar"},{"key":"ref47","first-page":"6131","article-title":"Sunrise: A simple unified framework for ensemble learning in deep reinforcement learning","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Lee"},{"key":"ref48","first-page":"2021","article-title":"Diagnosing bottlenecks in deep Q-learning algorithms","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Fu"},{"key":"ref49","first-page":"17604","article-title":"Regret minimization experience replay in off-policy reinforcement learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"34","author":"Liu"},{"key":"ref50","article-title":"Transient non-stationarity and generalisation in deep reinforcement learning","author":"Igl","year":"2020","journal-title":"arXiv:2006.05826"},{"key":"ref51","first-page":"11393","article-title":"Deep reinforcement learning amidst continual structured non-stationarity","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Xie"},{"key":"ref52","first-page":"4300","article-title":"Non-stationary reinforcement learning without prior knowledge: An optimal black-box approach","volume-title":"Proc. Conf. Learn. Theory","author":"Wei"},{"key":"ref53","first-page":"9976","article-title":"Non-stationary reinforcement learning under general function approximation","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Feng"},{"key":"ref54","first-page":"4572","article-title":"Variational policy gradient method for reinforcement learning with general utilities","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Zhang"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2023.110355"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM41043.2020.9155494"},{"key":"ref57","first-page":"7436","article-title":"Uncertainty-based offline reinforcement learning with diversified Q-ensemble","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"34","author":"An"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2020.12.023"},{"key":"ref59","first-page":"2635","article-title":"Learnability, stability and uniform convergence","volume":"11","author":"Shalev-Shwartz","year":"2010","journal-title":"J. Mach. Learn. Res."},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1007\/3-540-45014-9_1"},{"key":"ref61","first-page":"1","article-title":"On large-batch training for deep learning: Generalization gap and sharp minima","volume-title":"Proc. 5th Int. Conf. Learn. Represent.","author":"Keskar"},{"key":"ref62","first-page":"1","article-title":"Entropy-SGD: Biasing gradient descent into wide valleys","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Chaudhari"},{"key":"ref63","first-page":"1","article-title":"Formalizing Generalization and robustness of neural networks to weight perturbations","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Tsai"},{"key":"ref64","first-page":"1","article-title":"Intriguing properties of neural networks","volume-title":"Proc. 2nd Int. Conf. Learn. Represent.","author":"Szegedy"},{"key":"ref65","article-title":"Spectral norm regularization for improving the generalizability of deep learning","author":"Yoshida","year":"2017","journal-title":"arXiv:1705.10941"},{"key":"ref66","first-page":"1","article-title":"Bigger, regularized, optimistic: Scaling for compute and sample-efficient continuous control","volume-title":"Proc. 28th Annu. Conf. Neural Inf. Process. Syst.","author":"Nauman"},{"key":"ref67","first-page":"8787","article-title":"Decoupling value and policy for generalization in reinforcement learning","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Raileanu"}],"container-title":["IEEE Transactions on Systems, Man, and Cybernetics: Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/6221021\/11130474\/11053880.pdf?arnumber=11053880","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,20]],"date-time":"2025-08-20T06:17:06Z","timestamp":1755670626000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11053880\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9]]},"references-count":67,"journal-issue":{"issue":"9"},"URL":"https:\/\/doi.org\/10.1109\/tsmc.2025.3578050","relation":{},"ISSN":["2168-2216","2168-2232"],"issn-type":[{"value":"2168-2216","type":"print"},{"value":"2168-2232","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,9]]}}}