Geometry and Determinism of Optimal Stationary Control in Partially Observable Markov Decision Processes.
Guido MontúfarKeyan Ghazi-ZahediNihat AyPublished in: CoRR (2015)
Keyphrases
- partially observable markov decision processes
- optimal control
- dynamic programming
- finite state
- reinforcement learning
- planning under uncertainty
- belief space
- dynamical systems
- stochastic domains
- decision problems
- control policy
- belief state
- control system
- state space
- markov decision processes
- partially observable stochastic games
- average reward
- dec pomdps
- machine learning
- infinite horizon
- decision theoretic
- planning problems
- optimal policy
- special case
- computational complexity
- optimal solution
- multi agent