Reviews: Dynamic Incentive-Aware Learning: Robust Pricing in Contextual Auctions

The authors study the problem of setting (individual) reserve prices in a scenario of repeated contextual second-price auctions. The buyers are assumed strategic, i.e. they optimize a cumulative discounted utility, where their valuations are linear functions of the feature vector of a good. The considered scenario explicitly assumes existence of noise in the market. The seller's goal is to find an algorithm for setting prices that has sub-linear regret. Two algorithms are proposed: - the first one attain O(d log(Td) log(T)) regret bound, when the market noise distribution is known to the seller.