Partially Observable Markov Decision Processes

Concept

An extension of MDPs where the current state is not fully known, requiring agents to reason based on a history of observations and actions while managing uncertainty.

Mentioned in 1 video