A New Optimality Criterion for Nonhomogeneous Markov Decision Processes.
Wallace J. HoppJames C. BeanRobert L. SmithPublished in: Oper. Res. (1987)
Keyphrases
- optimality criterion
- markov decision processes
- average reward
- optimal policy
- state space
- policy iteration
- reinforcement learning
- risk sensitive
- finite state
- dynamic programming
- transition matrices
- average cost
- decision theoretic planning
- partially observable
- total reward
- reinforcement learning algorithms
- infinite horizon
- factored mdps
- state and action spaces
- discounted reward
- action space
- long run
- reward function
- function approximation
- evaluation function
- markov decision process
- action sets
- multistage