On Learning Action Costs from Input Plans
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913998823227392 |
|---|---|
| author | Morales, Marianela Pozanco, Alberto Canonaco, Giuseppe Gopalakrishnan, Sriram Borrajo, Daniel Veloso, Manuela |
| author_facet | Morales, Marianela Pozanco, Alberto Canonaco, Giuseppe Gopalakrishnan, Sriram Borrajo, Daniel Veloso, Manuela |
| contents | Most of the work on learning action models focus on learning the actions' dynamics from input plans. This allows us to specify the valid plans of a planning task. However, very little work focuses on learning action costs, which in turn allows us to rank the different plans. In this paper we introduce a new problem: that of learning the costs of a set of actions such that a set of input plans are optimal under the resulting planning model. To solve this problem we present $LACFIP^k$, an algorithm to learn action's costs from unlabeled input plans. We provide theoretical and empirical results showing how $LACFIP^k$ can successfully solve this task. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_10889 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | On Learning Action Costs from Input Plans Morales, Marianela Pozanco, Alberto Canonaco, Giuseppe Gopalakrishnan, Sriram Borrajo, Daniel Veloso, Manuela Artificial Intelligence Most of the work on learning action models focus on learning the actions' dynamics from input plans. This allows us to specify the valid plans of a planning task. However, very little work focuses on learning action costs, which in turn allows us to rank the different plans. In this paper we introduce a new problem: that of learning the costs of a set of actions such that a set of input plans are optimal under the resulting planning model. To solve this problem we present $LACFIP^k$, an algorithm to learn action's costs from unlabeled input plans. We provide theoretical and empirical results showing how $LACFIP^k$ can successfully solve this task. |
| title | On Learning Action Costs from Input Plans |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2408.10889 |