arXiv：cs.LG· Catherine Ji, Vivek Myers, Sergey Levine, Benjamin Eysenbach·· 4 小时前AI 评分35
赋权（Empowerment）的几何学：与技能学习方法的联系
The Geometry of Empowerment
AI 导读
该研究将赋权最大化与技能学习方法联系起来，为解读和分析赋权提供了新的几何视角，并回答了赋权与结构中心性之间关系的长期未解问题。分析还揭示了信息几何与奖励几何之间的区别，为构建可扩展的赋权最大化方法提供了理论启示。
正文
Abstract:Empowerment captures the capacity for an agent to actively control its environment. While conceptually appealing as an information-theoretic quantity, the connection between empowerment and structurally central states that provide broad access to future outcomes has remained an open question. In this work, we link empowerment maximization and skill-learning methods to provide new geometries for interpreting and analyzing empowerment. Our analyses answer longstanding open questions on the connections between empowerment and structural centrality. Our analyses also reveal distinctions between information and reward geometries, highlighting important theoretical implications to build scalable empowerment-maximization methods. Website and code can be found at this https URL.
| Comments: | 34 pages, 12 figures |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.07796 [cs.LG] |
| (or arXiv:2610.07796v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07796 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Catherine Ji [view email]
[v1]
Tue, 6 Oct 2026 05:45:19 UTC (1,906 KB)
来源：arXiv：cs.LG · arxiv.org