Goal-Conditioned Reinforcement Learning from Sub-Optimal Data on Metric Spaces
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910019068362752 |
|---|---|
| author | Reichlin, Alfredo Vasco, Miguel Yin, Hang Kragic, Danica |
| author_facet | Reichlin, Alfredo Vasco, Miguel Yin, Hang Kragic, Danica |
| contents | We study the problem of learning optimal behavior from sub-optimal datasets for goal-conditioned offline reinforcement learning under sparse rewards, invertible actions and deterministic transitions. To mitigate the effects of \emph{distribution shift}, we propose MetricRL, a method that combines metric learning for value function approximation with weighted imitation learning for policy estimation. MetricRL avoids conservative or behavior-cloning constraints, enabling effective learning even in severely sub-optimal regimes. We introduce distance monotonicity as a key property linking metric representations to optimality and design an objective that explicitly promotes it. Empirically, MetricRL consistently outperforms prior state-of-the-art goal-conditioned RL methods in recovering near-optimal behavior from sub-optimal offline data. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_10820 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Goal-Conditioned Reinforcement Learning from Sub-Optimal Data on Metric Spaces Reichlin, Alfredo Vasco, Miguel Yin, Hang Kragic, Danica Machine Learning We study the problem of learning optimal behavior from sub-optimal datasets for goal-conditioned offline reinforcement learning under sparse rewards, invertible actions and deterministic transitions. To mitigate the effects of \emph{distribution shift}, we propose MetricRL, a method that combines metric learning for value function approximation with weighted imitation learning for policy estimation. MetricRL avoids conservative or behavior-cloning constraints, enabling effective learning even in severely sub-optimal regimes. We introduce distance monotonicity as a key property linking metric representations to optimality and design an objective that explicitly promotes it. Empirically, MetricRL consistently outperforms prior state-of-the-art goal-conditioned RL methods in recovering near-optimal behavior from sub-optimal offline data. |
| title | Goal-Conditioned Reinforcement Learning from Sub-Optimal Data on Metric Spaces |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2402.10820 |