BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Physics-Integrated Hybrid Modeling  in Cardiac Digital Twins - Lin
 wei  Wang (Rochester Institute of Technology)
DTSTART:20240604T133000Z
DTEND:20240604T135000Z
UID:TALK214564@talks.cam.ac.uk
DESCRIPTION:The interest in leveraging physics-based inductive bias in dee
 p learning has resulted in recent developments of hybrid deep generative m
 odels (hybrid-DGMs) that integrates known physics-based mathematical expre
 ssions in neural generative models. The identification of these hybrid-DGM
 s often involves the inference of the parameters of the physics-based comp
 onent along with its neural component. The identfiability of these hybrid-
 DGM however has not yet been theoretically probed or established. While th
 e (un)identfiability of DGMs has been investigated\,&nbsp\; existing solut
 ions often do not apply here as they require observed auxiliary labels abo
 ut the underlying hybrid models. In this talk\, we probe the theoretical i
 dentfiability of hybrid-DGMs\, present meta-learning as a novel solution t
 o construct identifiable hybrid-DGMs\, and examine empirical evidence for 
 on synthetic and real-data benchmarks. We further discuss the possibility 
 to extend the identification of these hybrid-DGMs to unsupervised settings
  using high-dimensional observation data (e.g.\, image sequences)\, and it
 s proof-of-concept application to digital twins for cardiac electrophysiol
 ogy.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR