BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Diffusion modelling for amortised inference - Nikolay Malkin (Univ
 ersity of Edinburgh)
DTSTART:20250625T130000Z
DTEND:20250625T140000Z
UID:TALK232267@talks.cam.ac.uk
DESCRIPTION:This talk will survey recent work\, by me and others\, on the 
 use of diffusion models as amortised variational posteriors. While diffusi
 on models are classically trained to maximise a variational bound on datas
 et likelihood\, their expressive power can also be harnessed to approximat
 e posterior distributions over latent variables where no unbiased samples 
 are available &ndash\; that is\, amortised Bayesian inference &ndash\; and
  to approximately solve the related problem of sampling posteriors under d
 iffusion model priors. The ensuing learning problem has close connections 
 to stochastic optimal control and can be solved using a variety of learnin
 g-based and Monte Carlo approaches. After introducing these algorithms and
  connections\, I will present recent results on the use of techniques from
  deep reinforcement learning in diffusion sampling and on connections with
  (twisted) sequential Monte Carlo. Applications include high-dimensional i
 nverse problems in astrophysics and biology\, constrained sampling in larg
 e generative models\, inference of stochastic dynamical systems\, and blac
 k-box Bayesian optimisation.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR