BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:AI Meets Economics: The Case of Auction-Assisted AI Systems in Clo
 ud-Edge Continuum - Prof. Lei Jiao
DTSTART:20250523T140000Z
DTEND:20250523T150000Z
UID:TALK231856@talks.cam.ac.uk
CONTACT:Richard Mortier
DESCRIPTION:"Join via MS Teams":https://teams.microsoft.com/l/meetup-join/
 19%3ameeting_ZDE0NzFlMmYtODgyNi00N2M5LThhZGMtZGY2Zjk3M2FiOGJm%40thread.v2/
 0?context=%7b%22Tid%22%3a%2249a50445-bdfa-4b79-ade3-547b4f3986e9%22%2c%22O
 id%22%3a%22c74ff4ca-98fe-4b28-9889-e119acc12f30%22%7d\n\nMany cloud and ed
 ge AI services today perform machine learning inference in real time on en
 d user requests. Over time\, however\, models could degrade in accuracy du
 e to data and concept drifts\, and full retraining can be infeasible becau
 se of limited training data\, long training delay\, and prohibitive comput
 ational overhead. A promising solution is for the AI service to incorporat
 e externally supplied pre‑trained models to maintain resilience and accu
 racy in the face of evolving inputs. To incentivize third‑party model pr
 oviders\, who alone possess the requisite resources and data\, to produce 
 and contribute models\, an economic mechanism is required to monetize thei
 r contributions. Auction formats naturally suggest themselves\, yet they i
 ntroduce fundamental challenges in this circumstance: the interdependence 
 of sequential auctions\, the trade‑off between system overhead and infer
 ence performance\, and the need to balance economic properties with sustai
 ned 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 descr
 ibe the multiple performance guarantees of our approach\, including per‑
 auction truthfulness and individual rationality\, an upper bound on infere
 nce loss\, and a parameterized‑constant competitive ratio for social cos
 t\, all supported by empirical evaluations. Afterwards\, I will briefly su
 rvey our other efforts on auction‑assisted AI systems\, including edge A
 I inference over auctioned resources and foundation model fine‑tuning wi
 th auction‑based pricing. Finally\, I will conclude with a vision for fu
 ture research.\n\nBiography:\nLei Jiao received his Ph.D. in computer scie
 nce 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\, clou
 d/edge networks\, energy systems\, cybersecurity\, and multimedia. His wor
 k integrates mathematical methods from optimization\, control theory\, mac
 hine learning\, and economics. He has authored over 80 peer-reviewed publi
 cations in journals such as IEEE Transactions on Networking\, IEEE Transac
 tions on Mobile Computing\, IEEE Transactions on Parallel and Distributed 
 Systems\, and IEEE Journal on Selected Areas in Communications\, and in co
 nferences such as INFOCOM\, MOBIHOC\, ICDCS\, SECON\, ICNP\, ICPP\, and IP
 DPS\, garnering over 6\,000 citations according to Google Scholar. He is a
  recipient of the U.S. National Science Foundation CAREER Award\, the Ripp
 le Faculty Fellowship\, the Alcatel-Lucent Bell Labs UK and Ireland Recogn
 ition Award\, and several Best Paper Awards including those from IEEE CNS 
 2019 and IEEE LANMAN 2013. He has served in various program committee role
 s\, including as a track chair for ICDCS\, as a member for INFOCOM\, MOBIH
 OC\, ICDCS\, and WWW\, and as a chair for multiple workshops with INFOCOM 
 and ICDCS.\n\n
LOCATION:Computer Lab\, FW11 and Online (MS Teams link below)
END:VEVENT
END:VCALENDAR