BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Sparse and modular networks using exchangeable random measures - F
 rancois Caron (University of Oxford)
DTSTART:20160711T143000Z
DTEND:20160711T150000Z
UID:TALK66702@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Statistical network modeling has focused on representing      
          the graph as a discrete structure\, namely the adjacency         
       matrix\, and considering the exchangeability of this array.         
       In such cases\, it is well known that the graph is               nec
 essarily either dense (the number of edges scales               quadratica
 lly with the number of nodes) or trivially               empty. <br>      
          Here\, we instead consider representing the graph as a           
     measure on the plane. For the associated definition of               e
 xchangeability\, we rely on the Kallenberg representation               th
 eorem (Kallenberg\, 1990). For certain choices of such               excha
 ngeable random measures underlying the graph               construction\, 
 the network process is sparse with power-law               degree distribu
 tion\, and can accommodate an overlapping               block-structure. <
 br>               A Markov chain Monte Carlo algorithm is derived for     
           efficient exploration of the posterior distribution and allows t
 o recover the structure of a               range of networks ranging from 
 dense to sparse               based on our flexible formulation.<br>      
          <br>               Joint work with Emily Fox and Adrien Todeschin
 i<br>               <a href="http://arxiv.org/abs/1401.1137" target="_blan
 k" rel="nofollow">http://arxiv.org/abs/1401.1137</a><br>               <a 
 href="http://arxiv.org/pdf/1602.02114" target="_blank" rel="nofollow">http
 ://arxiv.org/pdf/1602.02114</a><br><br>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR