BEGIN:VCALENDAR VERSION:2.0 PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4// BEGIN:VEVENT UID:20251029T084921EDT-7751kzcf6R@132.216.98.100 DTSTAMP:20251029T124921Z DESCRIPTION:\n Mesoscale two-sample testing for networks\n\n  \n\n\nAbstract: \n\nNetworks arise naturally in many scientific fields as a representation of pairwise connections. Statistical network analysis has most often cons idered a single large network\, but it is common in a number of applicatio ns\, for example\, neuroimaging\, to observe multiple networks on a shared node set. When these networks are grouped by case-control status or anoth er categorical covariate\, the classical statistical question of two-sampl e comparison arises. In this work\, we address the problem of testing for statistically significant differences in a prespecified subset of the conn ections. This general framework allows an analyst to focus on a single nod e\, a specific region of interest\, or compare whole networks. In this “me soscale” setting\, we develop statistically sound projection-based tests f or two-sample comparison in both weighted and binary edge networks. Our ap proach can leverage all available network information\, and learn informat ive projections which improve testing power when low-dimensional network s tructure is present.\n\nSpeaker\n\nDr MacDonald is a postdoctoral scholar here at 9I. He received his PhD from the University of Mich igan in 2023\, under the supervision of Elizaveta Levina and Ji Zhu. Later this year\, Peter will begin his post as an Assistant Professor at the Un iversity of Waterloo.\n\nhttps://mcgill.zoom.us/j/87465663442\n\nMeeting I D: 874 6566 3442\n\nPasscode: None\n DTSTART:20240209T203000Z DTEND:20240209T213000Z LOCATION:Room 1104\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue Sherbrooke Ouest SUMMARY:Peter William MacDonald (9I) URL:/mathstat/channels/event/peter-william-macdonald-m cgill-university-355214 END:VEVENT END:VCALENDAR