BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:MRC Biostatistics Unit Seminars
SUMMARY:BSU Seminar: 'Double soft-thresholded model for mu
 lti-group scalar on vector-valued image regression
 ' - Prof Arkaprava Roy\, University of Florida
DTSTART;TZID=Europe/London:20221216T140000
DTEND;TZID=Europe/London:20221216T150000
UID:TALK193699AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/193699
DESCRIPTION:In this paper\, we develop a novel spatial variabl
 e selection method for scalar on vector-valued ima
 ge regression in a multi-group setting. Here\, ‘ve
 ctor-valued image’ refers to the imaging datasets 
 that contain vector-valued information at each pix
 el/voxel location\, such as in RGB color images\, 
 multimodal medical images\, DTI imaging\, etc. The
  focus of this work is to identify the spatial loc
 ations in the image having an important effect on 
 the scalar outcome measure. Specifically\, the ove
 rall effect of each voxel is of interest. We thus 
 develop a novel shrinkage prior by soft-thresholdi
 ng the ℓ2 norm of a latent multivariate Gaussian p
 rocess. It allows us to estimate sparse and piecew
 ise-smooth spatially varying vector-valued regress
 ion coefficient function. Motivated by the real da
 ta\, we further develop a double soft-thresholding
  based framework when there are multiple pre-speci
 fied subgroups. For posterior inference\, an effic
 ient MCMC algorithm is developed. We compute the p
 osterior contraction rate for parameter estimation
  and also establish consistency for variable selec
 tion of the proposed Bayesian model\, assuming tha
 t the true regression coefficients are Holder smoo
 th. Finally\, we demonstrate the advantages of the
  proposed method in simulation studies and further
  illustrate in an ADNI dataset for modeling MMSE s
 cores based on DTI-based vector-valued imaging mar
 kers.
LOCATION:Large Seminar Room\, East Forvie Building\, Forvie
  Site\, Robinson Way\, Cambridge CB2 0SR
CONTACT:Alison Quenault
END:VEVENT
END:VCALENDAR