For W at least C_d log(eT), minimax pseudo-regret in Lipschitz bandits is, up to logarithmic factors, the maximum of the sequential rate, a new memory-batch penalty T^((d+2)/(d+3)) (1+(B-1)W)^(-1/(d(d+3))), and a batch-depth term.
Proceedings of the Twenty-Third Conference on Uncertainty in Artificial Intelligence , pages=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Information Routing across Batch Boundaries: Memory--Batch Tradeoffs in Lipschitz Bandits
For W at least C_d log(eT), minimax pseudo-regret in Lipschitz bandits is, up to logarithmic factors, the maximum of the sequential rate, a new memory-batch penalty T^((d+2)/(d+3)) (1+(B-1)W)^(-1/(d(d+3))), and a batch-depth term.