BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Ensemble Inference Methods for Models with Noisy and Expensive Lik
 elihoods - Marie-Therese Wolfram (University of Warwick)
DTSTART:20220429T101500Z
DTEND:20220429T111500Z
UID:TALK171902@talks.cam.ac.uk
DESCRIPTION:The increasing availability of data presents an opportunity to
  calibrate unknown parameters which appear in complex models of phenomena 
 in the biomedical\, physical and social sciences. However\, model complexi
 ty often leads to parameter-to-data maps which are expensive to evaluate a
 nd are only available through noisy approximations. In this talk we focus 
 on interacting particle systems for the solution of the resulting inverse 
 problems for parameters. Of particular interest is the case where the avai
 lable forward model evaluations are subject to rapid fluctuations\, in par
 ameter space\, superimposed on the smoothly varying large scale parametric
  structure of interest. Multiscale analysis is then used to analyze the be
 haviour of interacting particle system algorithms when rapid fluctuations\
 , which we refer to as noise\, pollute the large scale parametric dependen
 ce of the parameter-to-data map. We compare ensemble Kalman methods and La
 ngevin-based methods in this light. The ensemble Kalman methods are shown 
 to behave favourably in the presence of noise in the parameter-to-data map
 \, whereas Langevin methods are adversely affected. On the other hand\, La
 ngevin methods have the correct equilibrium distribution in the setting of
  noise-free forward models\, whilst ensemble Kalman methods only provide a
 n uncontrolled approximation\, except in the linear case. We therefore int
 roduce a new class of algorithms - so called ensemble Gaussian process sam
 plers - which combine the benefits of both ensemble Kalman and Langevin me
 thods\, and perform favourably.\nJoint work with O.R.A. Dunbar\, A.B. Dunc
 an\, and A.M. Stuart
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR