BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Variational Flow Matching: A Tutorial on Generative AI for Scienti
 fic Applications - Jan-Willem van de Meent (Universiteit van Amsterdam)
DTSTART:20260212T110000Z
DTEND:20260212T120000Z
UID:TALK242344@talks.cam.ac.uk
DESCRIPTION:Flow matching has emerged as a powerful framework for generati
 ve modeling. It is finding increasing application in scientific domains\, 
 from molecular design to climate modeling. By directly regressing the velo
 city field of a deterministic ODE\, flow matching simplifies both training
  and sampling relative to diffusion-based approaches. This tutorial provid
 es a pedagogical introduction to flow matching through the lens of probabi
 listic inference. We will in particular discuss variational flow matching 
 (VFM)\, which reframes flow matching as self-supervised prediction of the 
 end point of a trajectory. This perspective enables extensions to discrete
  and structured domains &mdash\; essential for molecular and chemical syst
 ems &mdash\; as well as controlled generation via test-time guidance. The 
 tutorial will cover foundational concepts and implementation consideration
 s\, and will discuss how these methods can be applied to scientific challe
 nges involving both discrete objects (molecules\, proteins) and continuous
  fields (spatiotemporal signals in climate and biology).
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR