Talk on sudocodes - a fast compressed sensing reconstruction
approach that uses a hybrid dense/sparse measurement matrix
and two corresponding algorithms (joint work with I. Poltorak
and
D. Needell;
ppt).
Computational Information Processing
I am interested in application areas
where vast amounts of data must be processed effectively
and rapidly in order to extract useful - potentially "actionable" - information.
Specific such areas include
signal processing, communication systems,
distributed systems, networks, finance, medical imaging, and seismics.
To approach these problems, we must use a
multi-disciplinary approach, and combine tools
from information theory,
statistical signal processing, machine learning, and
computer science.
Specific research directions include: