方法对比

	半监督强化学习	域自适应强化学习
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2020s	2009–2020
提出者	Multiple contributors (Laskin, Srinivas, Abbeel et al.)	Multiple contributors (Taylor & Stone 2009 survey; Kim et al. 2020 among key formalizations)
类型	Semi-supervised training paradigm for RL agents	Transfer-based RL paradigm
开创性文献	Zhan, X., Zhu, X., & Shi, H. (2022). Deepthermal: Combustion optimization for thermal power generating units using offline reinforcement learning. Proceedings of the AAAI Conference on Artificial Intelligence, 36(4), 4680–4688.	Kim, K., Kim, H., Lim, H., & Choi, J. (2020). Domain Adaptive Reinforcement Learning with Model-Based Approach. arXiv preprint arXiv:2102.03170.
别名	SSRL, semi-supervised RL, RL with unlabeled data, label-efficient reinforcement learning	Domain-Adaptive RL, DARL, Cross-domain RL, Transfer RL with domain adaptation
相关	6	2
摘要	Semi-supervised reinforcement learning (SSRL) combines standard reinforcement learning — where an agent learns from sparse reward signals — with semi-supervised techniques that extract structure from unlabeled environment interactions. The goal is to improve sample efficiency and generalization when reward feedback is costly, delayed, or available only for a fraction of the agent's experience.	Domain-Adaptive Reinforcement Learning (DARL) extends standard RL by enabling a policy trained in one environment or domain to transfer and generalise effectively to a different but related target domain. It addresses the domain-shift problem — where dynamics, observations, or reward structures differ between training and deployment — through alignment, adaptation, or domain-randomisation techniques, reducing the need to collect costly experience in the target domain.