BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Extremal graphical models
DTSTART;VALUE=DATE-TIME:20221209T151500
DTEND;VALUE=DATE-TIME:20221209T170000
UID:0edc4886a5bb39249e68c5b5c34fbb659482aa47e242ee8a8e529389
CATEGORIES:Conferences - Seminars
DESCRIPTION:Sebastian Engelke (UniGE)\nEngelke and Hitz (2020\, JRSSB) int
roduce a new notion of conditional independence and graphical models for t
he most extreme observations of a multivariate sample. This enables the an
alysis of complex extreme events on network structures (e.g.\, floods) or
large-scale spatial data (e.g.\, heat waves). Recent results show that thi
s notion of extremal conditional independence arises as a special case of
a much more general theory for limits of sums and maxima of independent ra
ndom vectors.\nWe first discuss the implications of this theory on other f
ields\, and then focus on statistical inference for extremal graphical mod
els. This includes the estimation of model parameters on general graph str
uctures through matrix completion problems\, and data-driven structure lea
rning algorithms that estimate graphs through $L^1$ penalization. Theoreti
cal guarantees based on concentration inequalities are given even for high
-dimensional settings where the dimension $d$ is much larger than the samp
le size $n$. In extremes\, this is of particular interest since the effect
ive sample size $k$ is much smaller than $n$. \n
LOCATION:GA 3 21 https://plan.epfl.ch/?room==GA%203%2021
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR