ConceptAI & ML
Partial observability in policy gradients
Because the policy gradient derivation does not use the Markov property, it can be applied directly to partially observed Markov decision processes by replacing states with observations.
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.
Markov property1Partially observable Markov decision processes1
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.