Polarization phenomena: a new paradigm for probabilistic models
Speaker: Emmanuel Abbe
Series: Topical Seminars
Location: Engineering Quadrangle B205
Date/Time: Thursday, October 6, 2011, 4:30 p.m. - 5:30 p.m.
This talk starts with a brief description of the recently discovered polarization phenomenon. We review how this phenomenon can be used to address a problem opened more than sixty years ago: the construction of low-complexity codes provably capacity achieving. We then show how this phenomenon can be cast in a more general framework, beyond channel coding. Extensions to networks are overviewed, and emphasis is put on a new polarization phenomenon in a continuous setting. We discuss how this framework relates to signal acquisition problems, and how it connects to an open problem in compressed sensing: the construction of an efficient deterministic sensing matrix.
Emmanuel Abbe obtained his PhD from the EECS department at MIT, completing his thesis in the Laboratory for Information and Decision Systems. He previously obtained his B.S. and M.S. in the Mathematics Department at EPFL, receiving the CVCI Prize for his master thesis in statistics. He is currently a postdoctoral fellow and lecturer in the School of Communication and Computer Sciences at EPFL. His research interests are in information and coding theory, theoretical computer science and statistical signal processing. Emmanuel Abbe was recently awarded with the 2011 Latsis Prize, distinguishing a scientific work of particular excellence from all EPFL research areas.