Towards Using Fully Observable Policies for POMDPs.
András Attila SulyokKristóf KaracsPublished in: CoRR (2022)
Keyphrases
- partially observable markov decision processes
- fully observable
- finite state
- optimal policy
- reinforcement learning
- dynamical systems
- partial observability
- belief state
- planning under uncertainty
- decision problems
- dynamic programming
- markov decision problems
- state space
- planning problems
- belief space
- partially observable
- markov decision processes
- multi agent
- infinite horizon
- hidden state
- approximate solutions
- decentralized control
- initial state
- markov chain
- average cost
- np hard
- hidden markov models