Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
The Center for Control, Dynamical Systems, and Computation University of California at Santa Barbara
 

Summary: The Center for Control, Dynamical Systems, and Computation
University of California at Santa Barbara
Winter 2007 Seminar Series
Presents
A Variational View of Channel Coding
Sanjoy Mitter
Massachusetts Institute of Technology
Friday, March 16th, 2007 3:00pm-4:00pm ESB 2001
Abstract:
In the 60 or so years since Shannon's noisy channel coding theorem a number of papers have been
published on its connections with statistical mechanics. A common feature of these is that the phe-
nomenon of channel capacity is viewed as a type of phase transition.
The talk will develop these ideas, but in the framework of Bayesian in-ference. This admits a varia-
tional form, in which the posterior distribution minimises an information quantity we call the `apparent
information.' The variational formulation can be extended to infinite systems, where it shares
many of the properties of the Gibbs variational principle.
We shall consider a number of estimation problems associated with the noisy channel coding theo-
rem, and show that connections between the specific apparent information quantities in the different
layers of the communication architecture are at its heart. The specific apparent information quantities
for these problems exhibit a variety of phase transitions. We give preliminary results on the existence

  

Source: Akhmedov, Azer - Department of Mathematics, University of California at Santa Barbara

 

Collections: Mathematics