I will start with a basic question in coding theory: how to compress biased bits with a linear map. After revisiting the fundamental result of Shannon, I will discuss the challenge of achieving the limit with low-complexity matrices, and present two results using sparse and polar code matrices. I will then extend the results to multiple sources, and propose a framework for compression over the reals. This will take us to the problem of sensing high-dimensional real signals.
Recent developments in low-complexity coding
Emmanuel Abbe, Princeton University
Apr 25 2013 - 3:45pm
102A McDonnell Hall