DARLING augments RL with change detection to match minimax lower bounds on dynamic regret for piecewise stationary tabular and linear MDPs under separability and reachability conditions.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
DARLING: Detection Augmented Reinforcement Learning with Non-Stationary Guarantees
DARLING augments RL with change detection to match minimax lower bounds on dynamic regret for piecewise stationary tabular and linear MDPs under separability and reachability conditions.