CSL Master Calendar

Back to Listing

DCL Seminar: Yihong Wu - Coupling-based Converse Methods in Information Theory and Control

Event Type
Seminar/Symposium
Sponsor
Decision and Control Laboratory, Coordinated Science Laboratory
Location
CSL Auditorium, Room B02
Date
Mar 16, 2016   3:00 pm  
Speaker
Yihong Wu, Ph.D. University of Illinois
Contact
Linda Meccoli
E-Mail
lmeccoli@illinois.edu
Phone
217-333-9449
Views
45
Originating Calendar
CSL Decision and Control Group

Decision and Control Lecture Series

Coordinated Science Laboratory

 

“Coupling-based Converse Methods in Information Theory and Control”

 

Yihong Wu, Ph.D.

University of Illinois

 

Wednesday, March 16, 2016

3:00 p.m. to 4:00 p.m.

CSL Auditorium (B02)

____________________________________________________________________________________________________________________________________________________________________

 

“Coupling-based Converse Methods in Information Theory and Control” 

 

Abstract:

In this talk I will describe new coupling-based methods for proving impossibility results in information theory and stochastic control. With the common theme being bounding information measures by optimal transport (Wasserstein) distance, several applications will be discussed:

(a) optimal memoryless control on Gaussian line network consisting of n noisy Gaussian channels and n power-constrained relays, where we characterize the best non-linear contraction of total variation in Gaussian channels and show that the optimal end-to-end correlation decays as \Theta(\log\log n/\log n), resolving a problem of Lipsa-Martins (2011).

(b) the "missing corner-point" conjecture of Costa (1985) in Gaussian interference channel capacity region, where we settle this conjecture by showing that the output entropy is Lipschitz with respect to the Wasserstein distance and the unknown multi-user interference can be replaced by its iid approximations using couplings given by Talagrand's transportation-information inequality.

(c) strong data processing inequalities on Bayesian networks, where we bound the noisiness (in terms of Dobrushin's contraction coefficient) of the network in terms of those of the edges and the network topology.

and, if time permits,

(d) a simple proof of the Gaussian HWI inequalty of Otto-Villani.

This is based on joint work with Yury Polyanskiy (MIT).

 

Bio:

Yihong Wu received the B.E. degree from Tsinghua University, Beijing, China, in 2006 and the M.A. and Ph.D. degrees from Princeton University, Princeton, NJ, in 2008 and 2011, all in electrical engineering. Before joining the Department of Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign as an assistant professor in 2013, he was a postdoctoral fellow with the Statistics Department, The Wharton School, University of Pennsylvania, Philadelphia, PA. His research interests are in theoretical and algorithmic aspects of high-dimensional statistics and information theory.

link for robots only