Optimal Auctions through Deep Learning

Paul Duetting
Zhe Feng
David C. Parkes
Sai Srivatsa Ravindranath
Communications of the ACM, 64 (Issue 8)(2021), pp. 109-116


Designing an incentive compatible auction that maximizes expected revenue is an intricate task. The single-item case was resolved in a seminal piece of work by Myerson in 1981. Even after 30-40 years of intense research the problem remains unsolved for settings with two or more items. We overview recent research results that show how tools from deep learning are shaping up to be become a powerful tool for the automated design of optimal auctions. In this approach, an auction is modeled as a multi-layer neural network, with optimal auction design framed as a constrained learning problem, and solved through standard machine learning pipelines. Through this approach, it is possible to recover essentially all known analytical solutions for multi-item settings, and obtain novel mechanisms for settings in which the optimal mechanism is unknown.