ConceptAI & ML
Policy gradients under partial observability
Policy gradients can be adapted for partially observable markov decision processes by replacing state arguments with observations in the policy, but advantage estimators require careful handling since state-based value estimators rely on the Markov property.
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
1Connections
3The lectures that define it or assume it, and the concepts next to it.