Restarted Bayesian Online Change-point Detection for Non-Stationary Markov Decision Processes.
Réda AlamiMohammed MahfoudEric MoulinesPublished in: CoLLAs (2023)
Keyphrases
- markov decision processes
- change point detection
- non stationary
- change point
- finite state
- state space
- reinforcement learning
- optimal policy
- finite horizon
- normalized maximum likelihood
- outlier detection
- transition matrices
- policy iteration
- dynamic programming
- infinite horizon
- decision theoretic planning
- average cost
- average reward
- sequential data
- random fields
- markov decision process
- action sets
- action space
- stochastic shortest path
- multispectral