A Compromise Programming Approach to multiobjective Markov Decision Processes.
Wlodzimierz OgryczakPatrice PernyPaul WengPublished in: Int. J. Inf. Technol. Decis. Mak. (2013)
Keyphrases
- markov decision processes
- multi objective
- evolutionary algorithm
- state space
- finite state
- multi objective optimization
- optimal policy
- dynamic programming
- reinforcement learning
- multiple objectives
- planning under uncertainty
- transition matrices
- policy iteration
- average reward
- infinite horizon
- objective function
- average cost
- action sets
- reinforcement learning algorithms
- reachability analysis
- particle swarm optimization
- decision theoretic planning
- factored mdps
- genetic algorithm
- decision processes
- action space
- partially observable
- pareto optimal
- nsga ii
- markov decision process
- risk sensitive
- state and action spaces
- finite horizon
- least squares
- search algorithm
- semi markov decision processes
- neural network
- real time dynamic programming