Optimizing fixed-size stochastic controllers for POMDPs and decentralized POMDPs.
Christopher AmatoDaniel S. BernsteinShlomo ZilbersteinPublished in: Auton. Agents Multi Agent Syst. (2010)
Keyphrases
- fixed size
- reinforcement learning
- distributed constraint optimization
- dec pomdps
- partially observable markov decision processes
- belief state
- variable size
- continuous state
- partially observable
- point based value iteration
- markov decision processes
- sliding window
- multi agent
- decision theoretic
- dynamic programming
- finite state
- state space
- infinite horizon
- window size
- optimal policy
- small image patches
- policy search
- continuous state spaces
- belief space
- distributed systems
- monte carlo
- markov decision problems
- policy gradient
- control policies
- state dependent
- control system
- high quality