A Bayesian Clearing Mechanism for Combinatorial Auctions

  • 2018-11-16 16:38:03
  • Gianluca Brero, Sébastien Lahaie
  • 0

Abstract

We cast the problem of combinatorial auction design in a Bayesian frameworkin order to incorporate prior information into the auction process and minimizethe number of rounds to convergence. We first develop a generative model ofagent valuations and market prices such that clearing prices become maximum aposteriori estimates given observed agent valuations. This generative modelthen forms the basis of an auction process which alternates between refiningestimates of agent valuations and computing candidate clearing prices. Weprovide an implementation of the auction using assumed density filtering toestimate valuations and expectation maximization to compute prices. Anempirical evaluation over a range of valuation domains demonstrates that ourBayesian auction mechanism is highly competitive against the combinatorialclock auction in terms of rounds to convergence, even under the most favorablechoices of price increment for this baseline.

 

Quick Read (beta)

loading the full paper ...