Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.05031 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912878978662400 |
|---|---|
| author | Shehmar, Dikshant Schlegel, Matthew Taylor, Matthew E. Machado, Marlos C. |
| author_facet | Shehmar, Dikshant Schlegel, Matthew Taylor, Matthew E. Machado, Marlos C. |
| contents | Planning with a learned model remains a key challenge in model-based reinforcement learning (RL). In decision-time planning, state representations are critical as they must support local cost computation while preserving long-horizon structure. In this paper, we show that the Laplacian representation provides an effective latent space for planning by capturing state-space distances at multiple time scales. This representation preserves meaningful distances and naturally decomposes long-horizon problems into subgoals, also mitigating the compounding errors that arise over long prediction horizons. Building on these properties, we introduce ALPS, a hierarchical planning algorithm, and demonstrate that it outperforms commonly used baselines on a selection of offline goal-conditioned RL tasks from OGBench, a benchmark previously dominated by model-free methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_05031 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Laplacian Representations for Decision-Time Planning Shehmar, Dikshant Schlegel, Matthew Taylor, Matthew E. Machado, Marlos C. Machine Learning Planning with a learned model remains a key challenge in model-based reinforcement learning (RL). In decision-time planning, state representations are critical as they must support local cost computation while preserving long-horizon structure. In this paper, we show that the Laplacian representation provides an effective latent space for planning by capturing state-space distances at multiple time scales. This representation preserves meaningful distances and naturally decomposes long-horizon problems into subgoals, also mitigating the compounding errors that arise over long prediction horizons. Building on these properties, we introduce ALPS, a hierarchical planning algorithm, and demonstrate that it outperforms commonly used baselines on a selection of offline goal-conditioned RL tasks from OGBench, a benchmark previously dominated by model-free methods. |
| title | Laplacian Representations for Decision-Time Planning |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2602.05031 |