BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Statistical clustering of temporal networks through a dynamic stoc
hastic block model - Catherine Matias (CNRS (Centre national de la recherc
he scientifique)\; Université Pierre et Marie Curie Paris)
DTSTART:20161215T160000Z
DTEND:20161215T164500Z
UID:TALK69515@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION: Co-author: Vincent MIELE (CNRS / LBBE / Univ.
Lyon 1)
\;
Statistical node clustering in discr
ete time dynamic networks is an emerging field that raises many challenge
s. Here\, we explore statistical properties and frequentist inference in
a model that combines a stochastic block model (SBM) for its static part
with independent Markov chains for the evolution of the nodes groups thro
ugh time. We model binary data as well as weighted dynamic random graphs
(with discrete or continuous edges values). Our approach\,motivated by th
e importance of controlling for label switching issues across the differe
nt time steps\, focuses on detecting groups characterized by a stable wit
hin group connectivity behavior. We study identifiability of themodel par
ameters\, propose an inference procedure based on a variational expectati
on maximization algorithm as well as a model selection criterion to selec
t for the number of groups. We carefully discuss our initialization stra
tegy which plays an important role in the method and compare our procedur
e with exi sting ones on synthetic datasets. We also illustrate our approa
ch on dynamic contact networks\, one of encounters among high school stud
ents and two others on animal interactions. An implementation of the meth
od is available as a R package called dynsbm.
Related Li
nks
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR