BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Marius Zimand (Towson University)
DTSTART:20231206T160000Z
DTEND:20231206T171500Z
DTSTAMP:20260423T040000Z
UID:AAIT/26
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/AAIT/26/">Lo
 ssless expanders and Information Reconciliation with no shared randomness.
 </a>\nby Marius Zimand (Towson University) as part of Seminar on Algorithm
 ic Aspects of Information Theory\n\n\nAbstract\nLossless expanders are bip
 artite graphs in which for every sufficiently small set on the left side o
 f the bipartition\, most of the outgoing edges lead to distinct vertices. 
 I will present an application of such graphs in Information Reconciliation
 . This is a protocol with 2 parties: Sender and Receiver. Sender has a str
 ing x and Receiver has a string y. For instance\, think that x is an updat
 ed version of y. Sender sends a message m which allows Receiver to constru
 ct x. The goal is to minimize the length of m\, taking into account the co
 rrelation of x and y. This correlation can be formalized in different ways
  and is only known by Receiver. Standard solutions require Sender and Rece
 iver to share randomness. In my talk\, I will explain how lossless expande
 rs are used to obtain Information Reconciliation with no shared randomness
 .\n\nThe main idea is from the paper B. Bauwens and M. Zimand\, Universal 
 almost optimal compression and Slepian-Wolf coding in probabilistic polyno
 mial time\, Journal of the ACM\, April 2023 (also available at arXiv:1911.
 04268)\, but I'll focus on just one particular aspect.\n
LOCATION:https://researchseminars.org/talk/AAIT/26/
END:VEVENT
END:VCALENDAR
