BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Ensemble Methods in Machine Learning - Dr José Miguel Hernández 
 Lobato (University of Cambridge)
DTSTART:20120119T140000Z
DTEND:20120119T153000Z
UID:TALK35878@talks.cam.ac.uk
CONTACT:Konstantina Palla
DESCRIPTION:In this talk I will give an overview of the field of ensemble 
 methods.These techniques induce a collection (ensemble) of predictors and 
 then combine their output into a final consensus response which is usually
  more accurate than the output of the best individual predictor. The aggre
 gation process often generates a reduction in the bias and/or the variance
  components of the error of the final system. An important factor in the i
 mplementation of ensemble methods is to select the\noptimal size of the en
 semble. Over-estimation of this parameter can result in a waste of resourc
 es while under-estimation can result in loss of prediction accuracy. The l
 ast part of this talk will describe\na method for selecting the optimal si
 ze of a parallel ensemble of binary classifiers.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR