ConceptAI & ML
Empowerment
An intrinsic reward formulation that maximizes the mutual information between an agent's actions and outcomes to maximize control over the world.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Lecturers
2Each in their own terms, from the easiest level.
Sergey Levine
The mutual information between the next state and the current action, defined as I(sₜ₊₁; aₜ) = ℋ(sₜ₊₁) - ℋ(sₜ₊₁ | aₜ).
Connections
6The lectures that define it or assume it, and the concepts next to it.