Approximating Euclidean by Imprecise Markov Decision Processes.
Manfred JaegerGiorgio BacciGiovanni BacciKim Guldstrand LarsenPeter Gjøl JensenPublished in: CoRR (2020)
Keyphrases
- markov decision processes
- finite state
- optimal policy
- dynamic programming
- state space
- transition matrices
- finite horizon
- reinforcement learning
- policy iteration
- factored mdps
- euclidean space
- reinforcement learning algorithms
- planning under uncertainty
- reachability analysis
- decision theoretic planning
- markov decision process
- model based reinforcement learning
- decision processes
- average cost
- partially observable
- infinite horizon
- action sets
- average reward
- semi markov decision processes
- action space
- state abstraction
- reward function
- least squares
- np hard