BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Approximate Message Passing Algorithms - Dr Ramji Venkataramanan (
 University of Cambridge)
DTSTART:20160322T110000Z
DTEND:20160322T120000Z
UID:TALK65269@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:Approximate Message Passing” (AMP) refers to a class of iter
 ative algorithms that are Gaussian or quadratic approximations of loopy be
 lief propagation algorithms on dense factor graphs. AMP has attracted wide
 spread interest because it is significantly faster than traditional convex
  optimization procedures\, particularly for solving the classic \\ell_1-no
 rm based optimization that is typical in sparse signal recovery  (the LASS
 O).\n\nIn the first part of the talk\, I will discuss the main ideas behin
 d AMP with examples. In the second part\, I will describe an AMP decoding 
 algorithm for sparse regression codes\, a technique for communicating info
 rmation over Gaussian noise channels. In this setting\, the AMP decoder pr
 ovably achieves the optimal information-theoretic limit (the channel capac
 ity)\, and has excellent empirical performance as well.  I will conclude w
 ith some open questions about AMP and its connections to classical optimiz
 ation techniques.\n\nThis is joint work with Cynthia Rush and Adam Greig. 
 The talk will be self-contained and will not assume prior knowledge of mes
 sage passing/information theory/communications.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR