BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:The Unlearning Problem(s) - Anvith Thudi\, University of Toronto
DTSTART:20230620T130000Z
DTEND:20230620T140000Z
UID:TALK202510@talks.cam.ac.uk
CONTACT:Hridoy Sankar Dutta
DESCRIPTION:The talk presents challenges facing the study of machine unlea
 rning. The need for machine unlearning\, i.e.\, obtaining a model one woul
 d get without training on a subset of data\, arises from privacy legislati
 on and as a potential solution to data poisoning. The first part of the ta
 lk discusses approximate unlearning and the metrics one might want to stud
 y. We highlight methods for two desirable (though often disparate) notions
  of approximate unlearning. The second part departs from this line of work
  by asking if we can verify unlearning. Here we show how an entity can cla
 im plausible deniability\, and conclude that at the level of model weights
 \, being unlearnt is not always a well-defined property.\n\nRECORDING : Pl
 ease note\, this event will be recorded and will be available after the ev
 ent for an indeterminate period under a CC BY -NC-ND license. Audience mem
 bers should bear this in mind before joining the webinar or asking questio
 ns.\n
LOCATION:Webinar &amp\; LT2\, Computer Laboratory\, William Gates Building
 .
END:VEVENT
END:VCALENDAR