BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Budget-Limited Parametric Estimation for Expensive Black Box Engin
 eering Models - Andrew Duncan (Imperial College London)
DTSTART:20230426T090000Z
DTEND:20230426T100000Z
UID:TALK198424@talks.cam.ac.uk
DESCRIPTION:While parameter estimation methods for complex engineering mod
 els are well-established\, these tools often remain far out of the reach f
 or many engineering problems due to the computationally expensive nature o
 f the underlying models. The situation is typically far worse in settings 
 where we also wish to estimate associated posterior distribution via Marko
 v Chain Monte Carlo. In this work we present an approach which seeks to ov
 ercome this challenge by making judicious use of surrogate modelling and B
 ayesian experimental design. More specifically\, we investigate Bayesian p
 osterior inference in budget-limited settings where the likelihood can onl
 y be computed a fixed number of times. We reformulate posterior inference 
 as a sequential Bayesian experimental design problem. The 'unknown' likeli
 hood is characterised by a surrogate model\, which is sequentially updated
 \, along with with an ensemble of particles whose distribution will conver
 ge to the true posterior distribution. We show that this approach provides
  an effective approach to approximating the posterior distribution arising
  from computationally challenging Bayesian inverse problems\, and which ca
 n easily be extended to dynamic posterior updating in data-streaming scena
 rios relevant to digital twinning. We demonstrate the efficacy of this met
 hod on different benchmarks and compare with state-of-the-art methodology.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR