BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Reduced kernel rules for classification - Frederic Desobry\, Sigpr
 oc. Lab. CUED
DTSTART:20070329T120000Z
DTEND:20070329T130000Z
UID:TALK6698@talks.cam.ac.uk
CONTACT:Taylan Cemgil
DESCRIPTION:A new methodology is proposed to discriminate between two prob
 ability measures known through a set of data distributed according to eith
 er of these two measures. A decision rule is built as the plug-in of a\nke
 rnel rule\, defined on a small subset of the learning set. This methodolog
 y allows for fast yet accurate estimates of the optimal classification rul
 e. A statistical analysis yields consistency\nresults\, and rates of conve
 rgence for the probability of error. A dedicated model selection procedure
  is described\, and experiments illustrate further the comparison to state
 -of-the art classifiers.
LOCATION:LR5\, Engineering\, Department of
END:VEVENT
END:VCALENDAR