OOPE algorithm achieves regret scaling with effective dimension deff of offline data's Gram matrix and provides matching lower bounds in well- and poorly-explored regimes.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Regret minimization in Linear Bandits with offline data via extended D-optimal exploration
OOPE algorithm achieves regret scaling with effective dimension deff of offline data's Gram matrix and provides matching lower bounds in well- and poorly-explored regimes.