BEGIN:VCALENDAR VERSION:2.0 PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4// BEGIN:VEVENT UID:20250711T203254EDT-2459Umrku6@132.216.98.100 DTSTAMP:20250712T003254Z DESCRIPTION:\n Title: Multivariate Extremes Generator by Statistical Learnin g\n\n  \n\n Abstract:\n\n\nGenerating realistic extremes from an observation al dataset is crucial when trying to estimate the risks associated with th e occurrence of future extremes\, possibly of greater magnitude than those already observed. Generative approaches from the machine learning communi ty are not applicable to extreme samples without careful adaptation. On th e other hand\, asymptotic results from extreme value theory provide a theo retical framework for modeling multivariate extreme events\, through the n otion of multivariate regular variation. Bridging these two fields\, this presentation details a variational autoencoder approach for sampling multi variate distributions with heavy tails\, i.e.\, distributions likely to ex hibit extremes of particularly large intensities.\n\nSpeaker\n\nNicolas La fon is currently a postdoctoral researcher collaborating with Christian Ge nest and Johanna NeÅ¡lehová. He obtained his PhD from the Université Paris- Saclay at the Laboratory for Climate and Environmental Sciences in 2024 un der the supervision of Philippe Naveau and Ronan Fablet. His primary resea rch focuses on environmental extremes\, statistical learning\, and data as similation.\n\nhttps://mcgill.zoom.us/j/88929152266\n\nMeeting ID: 889 291 5 2266\n\nPasscode: None\n DTSTART:20250131T213000Z DTEND:20250131T223000Z LOCATION:Room 1104\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue Sherbrooke Ouest SUMMARY:Nicolas Lafon (ºÚÁϲ»´òìÈ University) URL:/mathstat/channels/event/nicolas-lafon-mcgill-univ ersity-362956 END:VEVENT END:VCALENDAR