BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Information Theory Seminar
SUMMARY:Generalization Bounds via Online Learning - Dr Ger
 gely Neu\, Universitat Pompeu Fabra
DTSTART;TZID=Europe/London:20230308T140000
DTEND;TZID=Europe/London:20230308T150000
UID:TALK194128AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/194128
DESCRIPTION:Bounding the generalization error is one most fund
 amental\nproblems in statistical learning theory. 
 In this talk\, I will present\na new framework for
  deriving generalization bounds from the\nperspect
 ive of online learning. Specifically\, we construc
 t an online\nlearning game called the Generalizati
 on Game\, where an online learner\nis trying to co
 mpete with a fixed statistical learning algorithm 
 in\npredicting the sequence of generalization gaps
  on a training set of\ni.i.d. data points. As I wi
 ll show\, this framework will allow us to\nrecover
  a range of classic bounds including PAC-Bayes and
 \ngeneralizations thereof. (Based on joint work wi
 th Gabor Lugosi.)
LOCATION:MR5\, CMS Pavilion A
CONTACT:Prof. Ramji Venkataramanan
END:VEVENT
END:VCALENDAR