On Minimizing Ordered Weighted Regrets in Multiobjective Markov Decision Processes.
Wlodzimierz OgryczakPatrice PernyPaul WengPublished in: ADT (2011)
Keyphrases
- markov decision processes
- multi objective
- finite state
- evolutionary algorithm
- state space
- transition matrices
- multi objective optimization
- reinforcement learning
- dynamic programming
- optimal policy
- decision theoretic planning
- policy iteration
- reachability analysis
- particle swarm optimization
- finite horizon
- planning under uncertainty
- decision processes
- multiple objectives
- infinite horizon
- reinforcement learning algorithms
- factored mdps
- risk sensitive
- objective function
- partially observable
- action space
- average cost
- pareto optimal
- reward function
- state abstraction
- state and action spaces
- action sets
- nsga ii
- model based reinforcement learning
- genetic algorithm
- markov decision process
- semi markov decision processes