Improving and Accelerating Offline RL in Large Discrete Action Spaces with Structured Policy Initialization
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908792986271744 |
|---|---|
| author | Landers, Matthew Killian, Taylor W. Hartvigsen, Thomas Doryab, Afsaneh |
| author_facet | Landers, Matthew Killian, Taylor W. Hartvigsen, Thomas Doryab, Afsaneh |
| contents | Reinforcement learning in discrete combinatorial action spaces requires searching over exponentially many joint actions to simultaneously select multiple sub-actions that form coherent combinations. Existing approaches either simplify policy learning by assuming independence across sub-actions, which often yields incoherent or invalid actions, or attempt to learn action structure and control jointly, which is slow and unstable. We introduce Structured Policy Initialization (SPIN), a two-stage framework that first pre-trains an Action Structure Model (ASM) to capture the manifold of valid actions, then freezes this representation and trains lightweight policy heads for control. On challenging discrete DM Control benchmarks, SPIN improves average return by up to 39% over the state of the art while reducing time to convergence by up to 12.8$\times$. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_04441 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Improving and Accelerating Offline RL in Large Discrete Action Spaces with Structured Policy Initialization Landers, Matthew Killian, Taylor W. Hartvigsen, Thomas Doryab, Afsaneh Machine Learning Reinforcement learning in discrete combinatorial action spaces requires searching over exponentially many joint actions to simultaneously select multiple sub-actions that form coherent combinations. Existing approaches either simplify policy learning by assuming independence across sub-actions, which often yields incoherent or invalid actions, or attempt to learn action structure and control jointly, which is slow and unstable. We introduce Structured Policy Initialization (SPIN), a two-stage framework that first pre-trains an Action Structure Model (ASM) to capture the manifold of valid actions, then freezes this representation and trains lightweight policy heads for control. On challenging discrete DM Control benchmarks, SPIN improves average return by up to 39% over the state of the art while reducing time to convergence by up to 12.8$\times$. |
| title | Improving and Accelerating Offline RL in Large Discrete Action Spaces with Structured Policy Initialization |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2601.04441 |