POMDP Filter: Pruning POMDP Value Functions with the Kaczmarz Iterative Method.
Eddy C. BoreraLarry D. PyeattArisoa S. RandrianasoloMahdi Naser-MoghadasiPublished in: MICAI (1) (2010)
Keyphrases
- reinforcement learning
- partially observable markov decision processes
- finite state
- partially observable
- partially observable markov decision process
- dynamical systems
- model free reinforcement learning
- belief state
- state space
- hidden state
- belief space
- markov decision process
- optimal policy
- continuous state
- dec pomdps
- soft constraints
- markov decision processes
- multi agent
- decision theoretic
- dynamic programming
- sufficient conditions
- multi objective
- infinite horizon
- learning algorithm
- planning problems
- noise reduction
- planning under uncertainty
- heuristic search
- partial observability
- policy evaluation
- hidden markov models
- computational complexity
- partially observable stochastic games