In first-price auctions with feedback-only shilling, an algorithm combining robust interval elimination and optimistic debiasing with racing achieves near-optimal regret rates of O(T^{2/3}) or O(sqrt(T)) and matches a lower bound in the single-active-region case.
Better Algorithms for Stochastic Bandits with Adversarial Corruptions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
We study the stochastic multi-armed bandits problem in the presence of adversarial corruption. We present a new algorithm for this problem whose regret is nearly optimal, substantially improving upon previous work. Our algorithm is agnostic to the level of adversarial contamination and can tolerate a significant amount of corruption with virtually no degradation in performance.
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Do Not Trust The Auctioneer: Learning to Bid in Feedback-Manipulated Auctions
In first-price auctions with feedback-only shilling, an algorithm combining robust interval elimination and optimistic debiasing with racing achieves near-optimal regret rates of O(T^{2/3}) or O(sqrt(T)) and matches a lower bound in the single-active-region case.