A Stopping Rule for Forecasting Horizons in Nonhomogeneous Markov Decision Processes.
James C. BeanWallace J. HoppIzak DuenyasPublished in: Oper. Res. (1992)
Keyphrases
- markov decision processes
- infinite horizon
- optimal policy
- finite state
- state space
- dynamic programming
- transition matrices
- partially observable
- reinforcement learning
- markov decision process
- policy iteration
- decision theoretic planning
- reachability analysis
- average cost
- decision processes
- reinforcement learning algorithms
- finite horizon
- factored mdps
- state and action spaces
- planning under uncertainty
- reward function
- average reward
- risk sensitive
- real time dynamic programming
- rule learning
- state abstraction
- probabilistic planning
- action space
- decision problems
- search algorithm