Parameter-free near-optimal regret bounds for non-stationary infinite-armed bandits are achieved via a blackbox restart scheme and a randomized elimination algorithm that tracks only significant rotting shifts.
ISBN 9781450355599
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Tracking Most Significant Shifts in Infinite-Armed Bandits
Parameter-free near-optimal regret bounds for non-stationary infinite-armed bandits are achieved via a blackbox restart scheme and a randomized elimination algorithm that tracks only significant rotting shifts.