BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20250812T191137EDT-6833EDGNvg@132.216.98.100
DTSTAMP:20250812T231137Z
DESCRIPTION: \n\nColloque des sciences mathématiques du Québec\n\n \n\nTitl
 e: Structure learning for Extremal graphical models\n\nAbstract:Extremal g
 raphical models are sparse statistical models for multivariate extreme eve
 nts. The underlying graph encodes conditional independencies and enables a
  visual interpretation of the complex extremal dependence structure. For t
 he important case of tree models\, we provide a data-driven methodology fo
 r learning the graphical structure. We show that sample versions of the ex
 tremal correlation and a new summary statistic\, which we call the extrema
 l variogram\, can be used as weights for a minimum spanning tree to consis
 tently recover the true underlying tree. Remarkably\, this implies that ex
 tremal tree models can be learned in a completely non-parametric fashion b
 y using simple summary statistics and without the need to assume discrete 
 distributions\, existence of densities\, or parametric models for marginal
  or bivariate distributions. Extensions to more general graphs are also di
 scussed.\n\nZoom: https://umontreal.zoom.us/j/93983313215?pwd=clB6cUNsSjAv
 RmFMME1PblhkTUts...\n\nID de réunion : 939 8331 3215 \n\nCode secret : 096
 952\n
DTSTART:20220218T203000Z
DTEND:20220218T213000Z
SUMMARY:Stanislav Volgushev (University of Toronto)
URL:/mathstat/channels/event/stanislav-volgushev-unive
 rsity-toronto-337394
END:VEVENT
END:VCALENDAR