BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Challenges in Bayesian inference and reliability with large number
 s of uncertain parameters - Jan Koune\, TU Delft
DTSTART:20231117T160000Z
DTEND:20231117T170000Z
UID:TALK206086@talks.cam.ac.uk
CONTACT:46601
DESCRIPTION:Efficient sampling methods play a central role in Bayesian inf
 erence\, reliability analysis and machine learning. Practical applications
  in structural and geotechnical engineering often deal with high-dimension
 al parameter spaces and computationally expensive physics-based models\, r
 endering many sampling-based approaches infeasible.\n\nIn this talk we ini
 tially examine these challenges through the case studies of a steel girder
  road bridge and a sheet pile quay wall. Motivated by recent work on machi
 ne learning approaches for nested sampling\, we explore the connection bet
 ween nested sampling and importance sampling\, subset simulation and norma
 lizing flows\, in an effort to derive an optimization-based approach for i
 nference and reliability estimation that takes advantage of the key ideas 
 behind nested sampling.
LOCATION:JDB Seminar Room\, CUED
END:VEVENT
END:VCALENDAR