A prior-independent revenue-maximizing auction for multiple additive bidders

Abstract

Recent work by Babaioff et al. [1], Yao [30], and Cai et al. [7] shows how to construct an approximately optimal auction for additive bidders, given access to the priors from which the bidders’ values are drawn. In this paper, building on the single sample approach of Dhangwatnotai et al. [15], we show how the auctioneer can obtain approximately optimal expected revenue in this setting without knowing the priors, as long as the item distributions are regular.