Optimal Rate Communication by Regression

SHARE:

Event Speaker: 

Andrew Barron (Yale University)

Event Location: 

32-155

Event Date/Time: 

Tuesday, April 8, 2014 - 4:00pm

Reception to follow.
 
Abstract
We discuss our recently developed sparse superposition codes for the Gaussian noise channel. With a fast adaptive successive decoder it achieves nearly exponentially small error probability at any fixed rate R less than the Shannon capacity. This is joint work with Antony Joseph and Sanghee Cho.
 
Biography
Professor Barron’s research interests include the areas of statistical information theory, statistical inference, model selection, optimal communications, the mininum description length principle, probability limit theorems, asymptotics of Bayes procedures, curve and surface estimation, artificial neural networks, approximation theory, and investment theory.  Received Ph.D., Electrical Engineering, Stanford University; M.S., Electrical Engineering, Stanford University; B.S. E.E. and Math Science, Rice University.  From 1985 - 1992 Andrew was Assistant and then Associate Professor of Statistics and Electrical & Computer Engineering, University of Illinois. From 1992 to present Andrew is a Professor of Statistics at Yale and has served terms as department chair, director of graduate studies, director of undergraduate studies in Statistics and director of undergraduate studies in Applied Mathematics.