Contextual Latent World Models for Offline Meta Reinforcement Learning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911482253410304 |
|---|---|
| author | Nakheai, Mohammadreza Scannell, Aidan Luck, Kevin Pajarinen, Joni |
| author_facet | Nakheai, Mohammadreza Scannell, Aidan Luck, Kevin Pajarinen, Joni |
| contents | Offline meta-reinforcement learning seeks to learn policies that generalize across related tasks from fixed datasets. Context-based methods infer a task representation from transition histories, but learning effective task representations without supervision remains a challenge. In parallel, latent world models have demonstrated strong self-supervised representation learning through temporal consistency. We introduce contextual latent world models, which condition latent world models on inferred task representations and train them jointly with the context encoder. This enforces task-conditioned temporal consistency, yielding task representations that capture task-dependent dynamics rather than merely discriminating between tasks. Our method learns more expressive task representations and significantly improves generalization to unseen tasks across MuJoCo, Contextual-DeepMind Control, and Meta-World benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_02935 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Contextual Latent World Models for Offline Meta Reinforcement Learning Nakheai, Mohammadreza Scannell, Aidan Luck, Kevin Pajarinen, Joni Machine Learning Offline meta-reinforcement learning seeks to learn policies that generalize across related tasks from fixed datasets. Context-based methods infer a task representation from transition histories, but learning effective task representations without supervision remains a challenge. In parallel, latent world models have demonstrated strong self-supervised representation learning through temporal consistency. We introduce contextual latent world models, which condition latent world models on inferred task representations and train them jointly with the context encoder. This enforces task-conditioned temporal consistency, yielding task representations that capture task-dependent dynamics rather than merely discriminating between tasks. Our method learns more expressive task representations and significantly improves generalization to unseen tasks across MuJoCo, Contextual-DeepMind Control, and Meta-World benchmarks. |
| title | Contextual Latent World Models for Offline Meta Reinforcement Learning |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2603.02935 |