Many cloud and edge AI services today perform machine learning inference in real time on end user requests. Over time, however, models could degrade in accuracy due to data and concept drifts, and full retraining can be infeasible because of limited training data, long training delay, and prohibitive computational overhead. A promising solution is for the AI service to incorporate externally supplied pre‑trained models to maintain resilience and accuracy in the face of evolving inputs. To incentivize third‑party model providers, who alone possess the requisite resources and data, to produce and contribute models, an economic mechanism is required to monetize their contributions. Auction formats naturally suggest themselves, yet they introduce fundamental challenges in this circumstance: the interdependence of sequential auctions, the trade‑off between system overhead and inference performance, and the need to balance economic properties with sustained participation. In this talk, firstly, I will formulate the repeated model‑procurement auctions as a non‑linear mixed‑integer social cost minimization problem, design a suite of polynomial‑time approximation algorithms that jointly solve this problem in an online manner, and describe the multiple performance guarantees of our approach, including per‑auction truthfulness and individual rationality, an upper bound on inference loss, and a parameterized‑constant competitive ratio for social cost, all supported by empirical evaluations. Afterwards, I will briefly survey our other efforts on auction‑assisted AI systems, including edge AI inference over auctioned resources and foundation model fine‑tuning with auction‑based pricing. Finally, I will conclude with a vision for future research.
Biography:
Lei Jiao received his Ph.D. in computer science from the University of Göttingen, Germany, in 2014. He is currently a faculty member at the University of Oregon, USA, and was previously a member of the technical staff at Nokia Bell Labs, Ireland. He researches networking and distributed computing, spanning AI infrastructures, cloud/edge networks, energy systems, cybersecurity, and multimedia. His work integrates mathematical methods from optimization, control theory, machine learning, and economics. He has authored over 80 peer-reviewed publications in journals such as IEEE Transactions on Networking, IEEE Transactions on Mobile Computing, IEEE Transactions on Parallel and Distributed Systems, and IEEE Journal on Selected Areas in Communications, and in conferences such as INFOCOM, MOBIHOC, ICDCS, SECON, ICNP, ICPP, and IPDPS, garnering over 6,000 citations according to Google Scholar. He is a recipient of the U.S. National Science Foundation CAREER Award, the Ripple Faculty Fellowship, the Alcatel-Lucent Bell Labs UK and Ireland Recognition Award, and several Best Paper Awards including those from IEEE CNS 2019 and IEEE LANMAN 2013. He has served in various program committee roles, including as a track chair for ICDCS, as a member for INFOCOM, MOBIHOC, ICDCS, and WWW, and as a chair for multiple workshops with INFOCOM and ICDCS.
