Course Blog‎ > ‎

... And on 1/29

posted Jan 28, 2014, 1:28 PM by Jen Mankoff
                           Department of Statistics
                          Carnegie Mellon University

                                  SEMINAR

                                 John Duchi
                              Ph.D. Candidate
                  Electrical Engineering & Computer Science
                    University of California, Berkeley


TITLE: Computation, Communication, and Privacy Constraints
       on Statistical Estimation

DATE AND TIME: Wednesday, Jan. 29, 2014, 4:00 - 5:00 PM

PLACE: Scaife Hall 125


                                Abstract

How do we integrate pratical considerations -- such as computation,
privacy, or communication -- into a rigorous theory of estimation?
In this talk, I discuss bringing these criteria in as constraints
in a minimax analysis of estimation, combining classical
decision-theoretic ideas with techniques from optimization and
information theory to understand the fundamental difficulties of
the problems.  By developing procedures that attain these lower
bounds, we obtain estimators that trade against a variety of
criteria for performance.  I will conclude with real-world
applications and examples in which our procedures exhibit good
practical performance.
_______________________________________________


Refreshments will be served in the Department of Statistics Lounge,
Baker Hall 132Q, at 3:30 p.m. preceding the seminar.


OPEN TO THE PUBLIC

For more information about Statistics Department seminars, see
http://www.stat.cmu.edu/seminar
Comments