Discounted deterministic Markov decision processes and discounted all-pairs shortest paths.
Omid MadaniMikkel ThorupUri ZwickPublished in: ACM Trans. Algorithms (2010)
Keyphrases
- markov decision processes
- optimal policy
- average reward
- state space
- infinite horizon
- stationary policies
- finite state
- reinforcement learning
- finite horizon
- dynamic programming
- average cost
- markov decision process
- transition matrices
- factored mdps
- policy iteration
- reinforcement learning algorithms
- partially observable
- model based reinforcement learning
- decision theoretic planning
- planning under uncertainty
- reachability analysis
- discounted reward
- decision processes
- risk sensitive
- action sets
- heuristic search
- search algorithm
- machine learning
- state and action spaces
- decision making
- decentralized control
- state abstraction
- decision diagrams
- action space
- dynamical systems