BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:A geometric approach for multivariate extremal inference
DTSTART:20241115T140000
DTEND:20241115T150000
DTSTAMP:20260510T140347Z
UID:926a16aa724c9decb7f38651966df20edee5459b680415bddf8ca9a1
CATEGORIES:Conferences - Seminars
DESCRIPTION:Ryan Campbell\, Lancaster University\nUntil recently\, differe
 nt types of joint tail dependence of random vectors required different mod
 elling procedures\, resulting in a lack of a unified approach for modellin
 g multivariate extremes. A new development remedies this by using the geom
 etry of the dataset to perform inference on the multivariate tail. A key q
 uantity in this inference is the so-called "gauge function"\, whose values
  define this geometry.\nIn this talk\, I'll present two methods to estimat
 e the gauge function given data. The first relies on parametric assumption
 s on the form of the gauge function. The second is semi-parametric\, inter
 polating the domain of the gauge function in a piecewise-linear fashion. T
 his results in a simple construction that is flexible on data with extrema
 l dependence behaviour that is difficult to parameterise\, and is more sui
 table for higher-dimensional applications. The piecewise-linear gauge func
 tion can be useful in defining a radial and an angular model\, allowing fo
 r the joint fitting of extremal pseudo-polar coordinates. This new methodo
 logy is applied to environmental datasets\, a setting where classical mult
 ivariate extremes methods often struggle due to the potential combination 
 of dependence and independence in the joint tails.\n\nJoint work with my P
 hD supervisor\, Jennifer Wadsworth.
LOCATION:CM 1 517 https://plan.epfl.ch/?room==CM%201%20517
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR