BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:5532@i2m.univ-amu.fr
DTSTART;TZID=Europe/Paris:20131216T105500
DTEND;TZID=Europe/Paris:20131216T115500
DTSTAMP:20241030T094418Z
URL:https://www.i2m.univ-amu.fr/evenements/friiam-frumam-e-morvant-ist-aus
 tria-domain-adaptation-of-weighted-majority-votes-via-perturbed-variation-
 based-self-labeling/
SUMMARY: (...): [Friiam/Frumam] E. Morvant (IST Austria): Domain Adaptation
  of Weighted Majority Votes via Perturbed Variation-based Self-Labeling
DESCRIPTION:: Domain Adaptation of Weighted Majority Votes via Perturbed Va
 riation-based Self-Labeling\n\nBy Emilie Morvant\\\, IST Austria.\n\nTalk 
 at Friiam/Frumam\\\, 2e étage\\\, St Charles\n\nIn machine learning\\\, t
 he domain adaptation problem arrives when the test (target) and the train 
 (source) data are generated from different distributions. A key applied is
 sue is thus the design of algorithms able to gen- eralize on a new distrib
 ution\\\, for which we have no label information. We consider the specific
  PAC-Bayesian situation focused on learning classification models defined 
 as a weighted majority vote over a set of real- valued functions. In this 
 context\\\, we present PV-MinCq a new framework that general- izes a non-a
 daptative algorithm (MinCq). PV-MinCq follows the next principle. Jus- tif
 ied by a theoretical bound on the tar- get risk of the vote\\\, we provide
  to MinCq a target sample labeled thanks to a per- tubed variation-based s
 elf-labeling focalized on the regions where the source and target marginal
 s appear similar. We also study the influence of our self-labeling\\\, fro
 m which we deduce an original process for tuning the hy- perparameters. Ou
 r experiments show very promising results on a synthetic problem.
CATEGORIES:Séminaire,Signal et Apprentissage
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:STANDARD
DTSTART:20131027T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
END:VTIMEZONE
END:VCALENDAR