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:Selection of the Regularization Parameter in Graph
ical Models using Network Characteristics - Natali
a Bochkina (University of Edinburgh)
DTSTART;TZID=Europe/London:20160727T140000
DTEND;TZID=Europe/London:20160727T143000
UID:TALK66860AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/66860
DESCRIPTION:We study gene interaction networks using Gaussian
graphical models that represent the underlying gra
ph structure of conditional dependence between ran
dom variables determined by their partial correlat
ion or precision matrix. In a high dimensional set
ting\, the precision matrix is estimated using pen
alized likelihood by adding a penalization term wh
ich controls the amount of sparsity in the precisi
on matrix and totally characterizes the complexity
and structure of the graph. The most commonly use
d penalization term is the L1 norm of the precisio
n matrix scaled by the regularization parameter wh
ich determines the tradeoff between sparsity of th
e graph and fit to the data. We propose several pr
ocedures to select the regularization parameter in
the estimation of graphical models that focus on
recovering reliably the appropriate network charac
teristic of the graph\, and discuss their Bayesian
interpretation. Performance is illustrated on sim
ulated data as well as on colon tumor gene express
ion data. This work is extended to reconstructing
a differential network between two paired groups o
f samples. \; \;

This is
joint work with with Adria Caballe Mestres (Unive
rsity of Edinburgh and Biomathematics and Statisti
cs Scotland) and Claus Mayer (Biomathematics and S
tatistics Scotland).
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR