BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260721T161734EDT-7398zmLshb@132.216.98.100
DTSTAMP:20260721T201734Z
DESCRIPTION:Title: Efficient Label Shift Adaptation through the Lens of Sem
 iparametric Models.\n\n\n	Abstract:\n\n\nWe study the domain adaptation pro
 blem with label shift in this work. Under the label shift context\, the ma
 rginal distribution of the label varies across the training and testing da
 tasets\, while the conditional distribution of features given the label is
  the same. Traditional label shift adaptation methods either suffer from l
 arge estimation errors or require cumbersome post-prediction calibrations.
  To address these issues\, we first propose a moment-matching framework fo
 r adapting the label shift based on the geometry of the influence function
 . Under such a framework\, we propose a novel method named efficient label
  shift adaptation (ELSA)\, in which the adaptation weights can be estimate
 d by solving linear systems. Theoretically\, the ELSA estimator is root-n 
 consistent (n is the sample size of the source data) and asymptotically no
 rmal. Empirically\, we show that ELSA can achieve state-of-the-art estimat
 ion performances without post-prediction calibrations\, thus\, gaining com
 putational efficiency.\n\n\n	Speaker\n\n\nDr. Qinglong Tian is an Assistant
  Professor in the Department of Statistics and Actuarial Science at the Un
 iversity of Waterloo. He obtain his PhD degree at Iowa State University wi
 th Professor William Meeker. His research focuses on engineering statistic
 s\, reliability and statistical computing.\n\nHybrid: In person\n\nLocatio
 n: Burnside Hall 1205\n\nZoom:\n\nhttps://mcgill.zoom.us/j/83436686293?pwd
 =b0RmWmlXRXE3OWR6NlNIcWF5d0dJQT09\n\nMeeting ID: 834 3668 6293\n\nPasscode
 : 12345\n
DTSTART:20230210T200000Z
DTEND:20230210T210000Z
LOCATION:Room 1205\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue
  Sherbrooke Ouest
SUMMARY:Qinglong Tian
URL:https://www.mcgill.ca/mathstat/channels/event/qinglong-tian-346027
END:VEVENT
END:VCALENDAR
