An Heuristic for Multi-Dimensional Markov Decision Processes
From MaRDI portal
A heuristic procedure for infinite horizon multidimensional Markov decision processes when certain simple policies are easy to evaluate, and it is not necessary to evaluate the improved policies, is given. Bounds on the loss of optimality from such policies are given, which may be used to accept or reject any such derived policy.
Recommendations
- Cost rate heuristics for semi-Markov decision processes
- Markov decision processes with multidimensional action spaces
- Suboptimal policy determination for large-scale Markov decision processes. I: Description and bounds
- scientific article; zbMATH DE number 3873074
- Reward Revision for Discounted Markov Decision Problems
Cites work
Cited in
(9)- Markov decision processes with multidimensional action spaces
- Generalizing Markov decision processes to imprecise probabilities
- Sensitivity-based nested partitions for solving finite-horizon Markov decision processes
- Heuristics as Markov chains
- Cost rate heuristics for semi-Markov decision processes
- scientific article; zbMATH DE number 934462 (Why is no real title available?)
- Suboptimal policy determination for large-scale Markov decision processes. II: Implementation and numerical evaluation
- Heuristic algorithm for nested Markov decision process: solution quality and computational complexity
- A pseudo-polynomial heuristic for path-constrained discrete-time Markovian-target search
This page was built for publication: An Heuristic for Multi-Dimensional Markov Decision Processes
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5286928)