ConceptAI & ML
Natural policy gradient
A policy gradient method that multiplies the regular gradient by the inverse of the Fisher Information Matrix to correctly account for parameter sensitivities.
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.
Opens up
Concepts that name this one as a prerequisite.
Lecturers
2Each in their own terms, from the easiest level.
Sergey Levine
David Silver
A parametrisation-independent gradient ascent direction that finds the steepest ascent step measured using the Fisher information matrix.
Connections
8The lectures that define it or assume it, and the concepts next to it.