BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:Multiple change-point estimation in high-dimension
al Gaussian graphical models - Sandipan Roy (Unive
rsity College London)
DTSTART;TZID=Europe/London:20180503T110000
DTEND;TZID=Europe/London:20180503T120000
UID:TALK104899AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/104899
DESCRIPTION:We consider the consistency properties of a regula
rised estimator for the simultaneous identificatio
n of both changepoints and graphical dependency st
ructure in multivariate time-series. Traditionally
\, estimation of Gaussian Graphical Models (GGM) i
s performed in an i.i.d setting. More recently\, s
uch models have been extended to allow for changes
in the distribution\, but only where changepoints
are known a-priori. In this work\, we study the G
roup-Fused Graphical Lasso (GFGL) which penalises
partial-correlations with an L1 penalty while simu
ltaneously inducing block-wise smoothness over tim
e to detect multiple changepoints. We present a pr
oof of consistency for the estimator\, both in ter
ms of changepoints\, and the structure of the grap
hical models in each segment. Several synthetic ex
periments and a real data application validate the
performance of the proposed methodology.
<
br>
LOCATION:Seminar Room 2\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR