BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Scalable sequential design for Bayesian inverse problems via condi
 tional transport - Karina Koval (Universität Heidelberg)
DTSTART:20250828T133000Z
DTEND:20250828T140000Z
UID:TALK234520@talks.cam.ac.uk
DESCRIPTION:We present a scalable approach to sequential optimal experimen
 tal design for Bayesian inverse problems with expensive forward models and
  high-dimensional parameters. By combining transport maps\, a derivative-b
 ased upper bound on expected information gain\, and dimension reduction vi
 a likelihood-informed subspaces\, our method enables tractable experimenta
 l design in a sequential setting. We demonstrate the effectiveness of the 
 approach with examples from groundwater flow and photoacoustic imaging.Thi
 s talk is based on joint work with Tiangang Cui\, Roland Herzog\, and Robe
 rt Scheichl.&nbsp\;
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR