A Lazy Abstraction Algorithm for Markov Decision Processes: Theory and Initial Evaluation.
Dániel SzekeresKristóf MarussyIstván MajzikPublished in: CoRR (2024)
Keyphrases
- markov decision processes
- dynamic programming
- learning algorithm
- objective function
- average reward
- model based reinforcement learning
- monte carlo
- np hard
- search space
- computational complexity
- optimal solution
- optimal policy
- reinforcement learning
- decision theoretic planning
- machine learning
- least squares
- sufficient conditions
- finite state
- reinforcement learning algorithms
- risk sensitive