Learning While Bidding in Vickrey Auctions with Acquisition Rate and Targeting Constraints

Explore a comprehensive lecture by Prof. Ravi Mazumdar from the University of Waterloo focusing on the theory and implementation of bidding strategies in Vickrey Auctions for digital advertising. Dive into the mechanics of ad placement in web browsing and mobile applications, where the market exceeds $100 billion. Learn about Demand Side Platforms (DSPs), real-time bidding constraints, and the development of least-cost bidding strategies in second-price auction contexts. Examine the rich primal-dual theory that emerges in both impressions and contracts spaces, along with the implementation of stochastic approximation algorithms for learning optimal bidding strategies with censored data. Understand how these algorithms solve constrained optimization problems without explicit information state learning, demonstrated through simulation results. The lecture, based on collaborative research with Ryan Kinnear, Amirhossein Boreiri, and Peter Marbach, was funded by Addictive Mobility Inc., Ontario OCE VIP II, and NSERC, addressing real-world challenges in digital advertising optimization.