BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Fully Stochastic Reconstruction Methods in Coupled Physics Imaging
  - Simon Arridge (University College London)
DTSTART:20230331T104000Z
DTEND:20230331T113000Z
UID:TALK198289@talks.cam.ac.uk
DESCRIPTION:Coupled Physics Imaging methods combine image contrast from on
 e physical process with observations using a secondary process\; several m
 odalities in acousto-optical imaging&nbsp\; follow this concept wherein op
 tical contrast is observed with acoustic measurements. For the inverse pro
 blem both an optical and acoustic model need to be inverted. Classical met
 hods that involve a non-linear optimisation approach can be combined with 
 advances in stochastic subsamplings strategies that are in part inspired b
 y machine learning applications. In such approaches the forward problem is
  considered deterministic and the stochasticity involves splitting of an o
 bjective function into sub functions that approach the fully sampled probl
 em in an expectation sense.\nIn this work we consider where the forward pr
 oblem is also solved stochastically\, by a Monte Carlo simulation of photo
 n propagation. By adjusting the batch size in the forward and inverse prob
 lems together\, we can achieve better performance than if subsampling is p
 erformed seperately.\n&nbsp\;\nJoint work with : S Powell\, C Macdonald\, 
 N H&auml\;nninen\, A Pulkkinen\, T Tarvainen
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR