Constraint-Adaptive Policy Switching for Offline Safe Reinforcement Learning

Authors: Yassine Chemingui, Aryan Deshwal, Honghao Wei, Alan Fern, Jana Doppa

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments on 38 tasks from the DSRL benchmark demonstrate that CAPS consistently outperforms existing methods, establishing a strong wrapper-based baseline for OSRL.
Researcher Affiliation Academia Yassine Chemingui1, Aryan Deshwal2, Honghao Wei1, Alan Fern3, Jana Doppa1 1Washington State University 2University of Minnesota 3Oregon State University EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes the algorithms and methods in prose and mathematical equations in Sections 4.1 and 4.2, but it does not include a distinct pseudocode block or algorithm figure.
Open Source Code Yes The code/appendices are available at https://github.com/ yassine Ch/CAPS.
Open Datasets Yes We employ 38 sequential decision-making benchmarks of varying difficulty from Safety-Gymnasium (Ray, Achiam, and Amodei 2019; Ji et al. 2024), Bullet Safety-Gym (Gronauer 2022), and Meta Drive (Li et al. 2022) within the DSRL framework (Liu et al. 2024). Further details are provided in Appendix C.1.
Dataset Splits No The paper mentions using a "fixed pre-collected dataset D" and evaluating algorithms with "three random seeds, and twenty episodes," but it does not specify explicit training, validation, or test splits for this dataset.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU models, CPU types, memory) used for running the experiments.
Software Dependencies No The paper mentions using existing offline RL methods like IQL and SAC+BC but does not specify software libraries, frameworks, or their version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes We provide the details of the neural network structure used for value and Q-functions, policy heads, and hyper-parameters in the Appendix C.3.