BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Maximum Mean Discrepancy and Variable Selection for High-Dimension
 al Data
DTSTART:20250325T130000Z
DTEND:20250325T140000Z
DTSTAMP:20260525T164700Z
UID:indico-event-14088@indico.math.cnrs.fr
DESCRIPTION:Speakers: Kensuke Mitsuzawa\n\nMaximum Mean Discrepancy (MMD) 
 [1] is a versatile metric for quantifying differences between probability 
 distributions\, with applications ranging from two-sample testing to gener
 ative model optimization. This presentation focuses on the problem of vari
 able selection: identifying the dimensions that contribute most significan
 tly to the dissimilarity between two distributions. Building upon MMD esti
 mator optimization [2]\, we introduce a regularization term that enables t
 he identification of these influential variables [3]. Our approach enhance
 s the interpretability of distributional comparisons by highlighting the k
 ey features driving observed differences. This methodology is demonstrated
  through empirical evaluations\, showcasing its effectiveness in discernin
 g relevant dimensions.\n \n[1] Gretton\, A.\, Borgwardt\, K. M.\, Rasch\,
  M. J.\, Schölkopf\, B.\, & Smola\, A. (2012). A kernel two-sample test. 
 The Journal of Machine Learning Research\, 13(1)\, 723-773.\n[2] Sutherlan
 d\, D. J.\, Tung\, H. Y.\, Strathmann\, H.\, De\, S.\, Ramdas\, A.\, Smola
 \, A.\, & Gretton\, A. (2016). Generative models and model criticism via o
 ptimized maximum mean discrepancy. arXiv preprint arXiv:1611.04488.\n[3] M
 itsuzawa\, K.\, Kanagawa\, M.\, Bortoli\, S.\, Grossi\, M.\, & Papotti\, P
 . (2023). Variable selection in maximum mean discrepancy for interpretable
  distribution comparison. arXiv preprint arXiv:2311.01537.\n\nhttps://indi
 co.math.cnrs.fr/event/14088/
LOCATION:Salle 1 (LJAD)
URL:https://indico.math.cnrs.fr/event/14088/
END:VEVENT
END:VCALENDAR