Abstract: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.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2508.11836 [cs.AI] |
| (or arXiv:2508.11836v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2508.11836 arXiv-issued DOI via DataCite |
From: Dave Goel [view email]
[v1]
Fri, 15 Aug 2025 23:05:37 UTC (2,110 KB)
[v2]
Thu, 21 May 2026 14:31:26 UTC (1,000 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。