Joint Learning of Hierarchical Neural Options and Abstract World Model
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_ | 1866916001475461120 |
|---|---|
| author | Piriyakulkij, Wasu Top Lehrach, Wolfgang Ellis, Kevin Murphy, Kevin |
| author_facet | Piriyakulkij, Wasu Top Lehrach, Wolfgang Ellis, Kevin Murphy, Kevin |
| contents | Building agents that can perform new skills by composing existing skills is a long-standing goal of AI agent research. Towards this end, we investigate how to efficiently acquire a sequence of skills, formalized as hierarchical neural options. However, existing model-free hierarchical reinforcement algorithms need a lot of data. We propose a novel method, which we call AgentOWL (Option and World model Learning Agent), that jointly learns -- in a sample efficient way -- an abstract world model (abstracting across both states and time) and a set of hierarchical neural options. We show, on a subset of Object-Centric Atari games, that our method can learn more skills using less data than baseline methods and possesses learning and generalization capabilities that the baselines do not have. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_02799 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Joint Learning of Hierarchical Neural Options and Abstract World Model Piriyakulkij, Wasu Top Lehrach, Wolfgang Ellis, Kevin Murphy, Kevin Machine Learning Artificial Intelligence Building agents that can perform new skills by composing existing skills is a long-standing goal of AI agent research. Towards this end, we investigate how to efficiently acquire a sequence of skills, formalized as hierarchical neural options. However, existing model-free hierarchical reinforcement algorithms need a lot of data. We propose a novel method, which we call AgentOWL (Option and World model Learning Agent), that jointly learns -- in a sample efficient way -- an abstract world model (abstracting across both states and time) and a set of hierarchical neural options. We show, on a subset of Object-Centric Atari games, that our method can learn more skills using less data than baseline methods and possesses learning and generalization capabilities that the baselines do not have. |
| title | Joint Learning of Hierarchical Neural Options and Abstract World Model |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2602.02799 |