Finite Automata Extraction: Low-data World Model Learning as Programs from Gameplay Video
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866918515910377472 |
|---|---|
| author | Goel, Dave Guzdial, Matthew Sarkar, Anurag |
| author_facet | Goel, Dave Guzdial, Matthew Sarkar, Anurag |
| contents | World models are defined as a compressed spatial and temporal learned representation of an environment. The learned representation is typically a neural network, making transfer of the learned environment dynamics and explainability a challenge. In this paper, we propose an approach, Finite Automata Extraction (FAE), that learns a neuro-symbolic world model from gameplay video represented as programs in a novel domain-specific language (DSL): Retro Coder. Compared to prior world model approaches, FAE learns a more precise model of the environment and more general code than prior DSL-based approaches. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_11836 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Finite Automata Extraction: Low-data World Model Learning as Programs from Gameplay Video Goel, Dave Guzdial, Matthew Sarkar, Anurag Artificial Intelligence World models are defined as a compressed spatial and temporal learned representation of an environment. The learned representation is typically a neural network, making transfer of the learned environment dynamics and explainability a challenge. In this paper, we propose an approach, Finite Automata Extraction (FAE), that learns a neuro-symbolic world model from gameplay video represented as programs in a novel domain-specific language (DSL): Retro Coder. Compared to prior world model approaches, FAE learns a more precise model of the environment and more general code than prior DSL-based approaches. |
| title | Finite Automata Extraction: Low-data World Model Learning as Programs from Gameplay Video |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2508.11836 |