BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A Bayesian nonparametric model for sparse dynamic networks - Konst
 antina Palla (University of Oxford)
DTSTART:20160726T110000Z
DTEND:20160726T113000Z
UID:TALK66850@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:<span>Co-authors: Francois Caron (Univ of OXford)\, Yee  Whye 
 Teh (Univ of Oxford) <br></span><span><br>We propose a Bayesian nonparamet
 ric prior for time-varying  networks. To each node of the network is assoc
 iated a positive parameter\,  modeling the sociability of that nodes. Soci
 abilities are assumed to evolve over  time\, and are modeled via a dynamic
  point process model. The model is able to  (a) capture smooth evolution o
 f the interactions between nodes\, allowing edges  to appear/disappear ove
 r time (b) capture long term evolution of the  sociabilities of the nodes 
 (c) and yields sparse graphs\, where the number of  edges grows subquadrat
 ically with the number of nodes. The evolution of the  sociabilities is de
 scribed by a tractable time-varying gamma process. We provide  some theore
 tical insights into the model\, describe a Hamiltonian Monte Carlo  algori
 thm for efficieent exploration of the posterior distribution and present  
 results on synthetic and real world dataset.</span>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR