Toward real/interactive-time for l<sup>1</sup> related problem

Wednesday, October 7, 2009 - 9:00am - 9:45am
Lind 305
Jérôme Darbon (École Normale Supérieure de Cachan)
Keywords: Parallel Programming, optimization, l1

Abstract: We consider the recovery of signal via compressive sensing where the signal
itself or its gradient are assumed to be sparse. This amounts to solve
a l1 or a Total Variation minimization problem.
We propose minimization algorithms specifically designed to take advantage
of shared memory, vectorized, parallel and manycore microprocessors such as
the Cell processor, new generation Graphics Processing Units (GPUs) and
standard vectorized multicore processors (e.g. standard quad core CPUs).

MSC Code: