BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Bayesian Optimization for Function-Valued Responses under Min–Ma
 x Criteria
DTSTART:20260219T140000Z
DTEND:20260219T150000Z
DTSTAMP:20260414T085600Z
UID:indico-event-85@conferences.enbis.org
DESCRIPTION:Speakers: Davide Cacciarelli (Imperial College London)\n\n\nBa
 yesian Optimization for Function-Valued Responses under Min–Max Criteria
 \nSpeaker: Pouya Ahadi\nChair: Davide Cacciarelli\nDate: 19th February 20
 26\, at 15:00-16:00 CET\nAbstract:\nBayesian optimization is widely used f
 or optimizing expensive black box functions\, but most existing approaches
  focus on scalar responses. In many scientific and engineering settings th
 e response is functional\, varying smoothly over an index such as time or 
 wavelength\, which makes classical formulations inadequate. Existing metho
 ds often minimize integrated error\, which captures average performance bu
 t neglects worst case deviations. To address this limitation we propose mi
 n-max Functional Bayesian Optimization (MM-FBO)\, a framework that directl
 y minimizes the maximum error across the functional domain. Functional res
 ponses are represented using functional principal component analysis\, and
  Gaussian process surrogates are constructed for the principal component s
 cores. Building on this representation\, MM-FBO introduces an integrated u
 ncertainty acquisition function that balances exploitation of worst case e
 xpected error with exploration across the functional domain. We provide tw
 o theoretical guarantees: a discretization bound for the worst case object
 ive\, and a consistency result showing that as the surrogate becomes accur
 ate and uncertainty vanishes\, the acquisition converges to the true min-m
 ax objective. We validate the method through experiments on synthetic benc
 hmarks and physics inspired case studies involving electromagnetic scatter
 ing by metaphotonic devices and vapor phase infiltration. Results show tha
 t MM-FBO consistently outperforms existing baselines and highlights the im
 portance of explicitly modeling functional uncertainty in Bayesian optimiz
 ation.\nBio:\nPouya Ahadi is a PhD candidate in Machine Learning at the Ge
 orgia Institute of Technology\, advised by Dr. Kamran Paynabar. His resear
 ch focuses on active learning\, Bayesian optimization\, functional data an
 alysis\, and robust methods for inverse problems and anomaly detection. Pr
 ior to his PhD\, he earned a Master’s degree in Industrial Engineering f
 rom Oklahoma State University and a Bachelor’s degree in Industrial Engi
 neering from Sharif University of Technology.\n\nhttps://conferences.enbis
 .org/event/85/
URL:https://conferences.enbis.org/event/85/
END:VEVENT
END:VCALENDAR