BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An efficient kernel product for automatic differentiation librarie
 s\, with applications to measure transport - Jean Feydy (École Normale Su
 périeure\; ENS de Cachan)
DTSTART:20171113T110000Z
DTEND:20171113T113000Z
UID:TALK95005@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Authors : Benjamin Charlier\, Jean Feydy\, Joan Alexis Glaun&e
 grave\;s and Alain Trouv&eacute\;  This paper presents a memory-efficient 
 implementation of the kernel matrix-vector product\, which is suitable for
  use with automatic differentiation libraries -- in our case\, PyTorch. Th
 is piece of software alleviates the major bottleneck of autodiff libraries
  as far as diffeomorphic image registration is concerned: symbolic python 
 code can now scale up to large point clouds and shapes (100\,000+ vertices
 ). To showcase the value of automatic differentiation to the LDDMM communi
 ty\, we introduce the "normalized Hamiltonian" setting and show that it co
 rresponds to a spatially regularized optimal transport of mass distributio
 ns: made tractable by autodiff libraries\, the kernel normalization trick 
 turns an extrinsic image deformation routine into an intrinsic measure tra
 nsportation program.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR