Localized Dynamics-Aware Domain Adaption for Off-Dynamics Offline Reinforcement Learning
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914347525079040 |
|---|---|
| author | Xia, Zhangjie Yang, Yu Xu, Pan |
| author_facet | Xia, Zhangjie Yang, Yu Xu, Pan |
| contents | Off-dynamics offline reinforcement learning (RL) aims to learn a policy for a target domain using limited target data and abundant source data collected under different transition dynamics. Existing methods typically address dynamics mismatch either globally over the state space or via pointwise data filtering; these approaches can miss localized cross-domain similarities or incur high computational cost. We propose Localized Dynamics-Aware Domain Adaptation (LoDADA), which exploits localized dynamics mismatch to better reuse source data. LoDADA clusters transitions from source and target datasets and estimates cluster-level dynamics discrepancy via domain discrimination. Source transitions from clusters with small discrepancy are retained, while those from clusters with large discrepancy are filtered out. This yields a fine-grained and scalable data selection strategy that avoids overly coarse global assumptions and expensive per-sample filtering. We provide theoretical insights and extensive experiments across environments with diverse global and local dynamics shifts. Results show that LoDADA consistently outperforms state-of-the-art off-dynamics offline RL methods by better leveraging localized distribution mismatch. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_21072 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Localized Dynamics-Aware Domain Adaption for Off-Dynamics Offline Reinforcement Learning Xia, Zhangjie Yang, Yu Xu, Pan Machine Learning Artificial Intelligence Robotics Off-dynamics offline reinforcement learning (RL) aims to learn a policy for a target domain using limited target data and abundant source data collected under different transition dynamics. Existing methods typically address dynamics mismatch either globally over the state space or via pointwise data filtering; these approaches can miss localized cross-domain similarities or incur high computational cost. We propose Localized Dynamics-Aware Domain Adaptation (LoDADA), which exploits localized dynamics mismatch to better reuse source data. LoDADA clusters transitions from source and target datasets and estimates cluster-level dynamics discrepancy via domain discrimination. Source transitions from clusters with small discrepancy are retained, while those from clusters with large discrepancy are filtered out. This yields a fine-grained and scalable data selection strategy that avoids overly coarse global assumptions and expensive per-sample filtering. We provide theoretical insights and extensive experiments across environments with diverse global and local dynamics shifts. Results show that LoDADA consistently outperforms state-of-the-art off-dynamics offline RL methods by better leveraging localized distribution mismatch. |
| title | Localized Dynamics-Aware Domain Adaption for Off-Dynamics Offline Reinforcement Learning |
| topic | Machine Learning Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2602.21072 |