Singular control for discounted Markov Decision Processes in a general state space.
Oswaldo L. V. CostaFrançois DufourPublished in: CDC/ECC (2011)
Keyphrases
- markov decision processes
- state space
- optimal policy
- reinforcement learning
- dynamic programming
- reinforcement learning algorithms
- factored mdps
- transition matrices
- infinite horizon
- decentralized control
- finite state
- average reward
- action space
- policy iteration
- markov decision process
- decision theoretic planning
- control problems
- model based reinforcement learning
- partially observable
- reachability analysis
- decision processes
- average cost
- control system
- state abstraction
- heuristic search
- markov chain
- reward function
- finite horizon
- planning problems
- planning under uncertainty
- markov decision problems
- action sets
- optimal control
- sufficient conditions
- dynamical systems
- state variables
- search algorithm