Smart sampling for lightweight verification of Markov decision processes.
Pedro R. D'ArgenioAxel LegaySean SedwardsLouis-Marie TraonouezPublished in: Int. J. Softw. Tools Technol. Transf. (2015)
Keyphrases
- lightweight
- markov decision processes
- finite state
- reinforcement learning
- state space
- dynamic programming
- optimal policy
- model checking
- transition matrices
- decision theoretic planning
- finite horizon
- reachability analysis
- reinforcement learning algorithms
- average reward
- risk sensitive
- planning under uncertainty
- action space
- policy iteration
- infinite horizon
- markov decision process
- state and action spaces
- model based reinforcement learning
- sample size
- reward function
- development environments
- decision processes
- average cost
- monte carlo
- factored mdps
- partially observable
- communication infrastructure
- action sets
- discounted reward
- real time dynamic programming
- function approximation
- state abstraction
- decision diagrams
- probabilistic planning