2020-12-23 | Tao Lin:Robustness of Empirical Revenue Maximization in Auction Learning
2020-12-23
Abstract
Empirical RevenueMaximization (ERM) is an important price learning algorithm in data-drivenauction design. It learns, from bidders value distribution, an approximatelyrevenue optimal reservation price in both repeated auctions and uniform-priceauctions. However, in these scenariosthe bidders who provide samples to ERM have incentives to manipulate thesamples in order to lower the output price. We show that ERM is robust against such manipulation, as long as the numberof manipulated samples is small. Specifically,we generalize a measure called “incentive-awareness measure”proposed by Lavi et al (2019) to quantify the reduction of ERM’s output due to a change of 1 ≤ m≤ o(N^0.5) out of N input samples, and provide specific convergencerates of this measure to zero as N goes to infinity. By adopting this measure, we use ERM toconstruct an efficient, approximately incentive-compatible, and revenue-optimallearning algorithm in repeated auctions against non-myopic bidders, and showapproximate group-IC in uniform-price auctions.This is joint work withXiaotie Deng, Ron Lavi, Qi Qi, Wenwei Wang, Xiang Yan, accepted by NeurIPS’20 (https://arxiv.org/abs/2010.05519)
Time
12月23日(星期三)14:00-15:00
Speaker
Tao Lin is a first-year PhDstudent at Harvard University, advised by Prof. Yiling Chen. He obtained his Bachelor’s degree in CS fromTuring Class, Peking University, where he worked under the supervision of Prof.Xiaotie Deng.
His research interests liein the intersection between economics and computer science, in particular,algorithmic game theory.
Venue
信管学院602室