BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Compressible distributions and Compressed Sensing - Prof Mike Davi
 es\, School of Engineering and Electronics\, University of Edinburgh
DTSTART:20111212T141500Z
DTEND:20111212T150000Z
UID:TALK35004@talks.cam.ac.uk
CONTACT:Rachel Fogg
DESCRIPTION:We consider the problem of compressed sensing when the signal 
 is drawn from a statistical signal model and identify probability distribu
 tions whose independent and identically distributed (iid) realizations are
  compressible or incompressible\, i.e.\, can/cannot be approximated as spa
 rse. Within this setting we consider some sample-distortion functions for 
 i.i.d. distributions and derive a simple sample distortion lower bound. Vi
 a this we will argue that Lapace distribution associated with the MAP inte
 rpretation of the popular L1 reconstruction algorithm is really not compre
 ssible.\nWe then extend the compressible model to consider a stochastic mu
 lti-resolution image model. Using empirical sample distortion functions we
  are able to compute an optimal bandwise sampling strategy and to accurate
 ly predict the compressed sensing possible performance gains available in 
 compressive imaging.\n
LOCATION:LR5\, Engineering\, Department of
END:VEVENT
END:VCALENDAR