On Anderson acceleration for partially observable Markov decision processes.
Melike ErmisMingyu ParkInsoon YangPublished in: CoRR (2021)
Keyphrases
- partially observable markov decision processes
- finite state
- reinforcement learning
- planning under uncertainty
- decision problems
- dynamical systems
- belief state
- optimal policy
- belief space
- continuous state
- dynamic programming
- partially observable stochastic games
- state space
- partial observability
- stochastic domains
- markov decision processes
- planning problems
- partially observable
- partially observable domains
- sequential decision making problems
- multi agent
- approximate solutions
- partially observable markov decision process
- partially observable markov
- infinite horizon
- markov chain
- np hard
- initial state
- data mining
- dec pomdps
- domain specific
- general purpose
- point based value iteration
- bayesian networks
- decision trees