The authors distinguish potential empowerment, the maximum information an agent’s actions could convey about its future state, from effective empowerment, the information actually transmitted by a policy or set of skills.
Empowerment Leads Agents to Central States, but Not Always Adaptability
The authors prove that empowerment corresponds to structural centrality in tabular environments, while showing that this link can fail for continuous states or uneven reward distributions.
Top University
Catherine Ji · Vivek Myers · Sergey Levine · Benjamin Eysenbach
Princeton University · UC Berkeley
Research Digest··2 min read
Ji and colleagues give geometric and temporal interpretations of empowerment, an information-theoretic measure of how strongly an agent can control its future state.
Why this paper
From Princeton University and UC Berkeley
In one line
Empowered states are central in information geometry, enabling task adaptation only in tabular settings.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (4 noted)
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§