BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Classification with unknown class conditional label noise on non-c
 ompact feature spaces - Henry Reeve — University of Birmingham
DTSTART:20191101T140000Z
DTEND:20191101T150000Z
UID:TALK130057@talks.cam.ac.uk
CONTACT:Dr Sergio Bacallado
DESCRIPTION:We consider the problem of classification in the presence of l
 abel noise.  In the analysis of classification problems it is typically as
 sumed that the train and test distributions are one and the same. In pract
 ice\, however\, it is often the case that the labels in the training data 
 have been corrupted with some unknown probability. We shall focus on class
 ification with class conditional label noise in which the labels observed 
 by the learner have been corrupted with some unknown probability which is 
 determined by the true class label.\n\nIn order to obtain finite sample ra
 tes\, previous approaches to classification with unknown class conditional
  label noise have required that the regression function attains its extrem
 a uniformly on sets of positive measure. We consider this problem in the s
 etting of non-compact metric spaces\, where the regression function need n
 ot attain its extrema.\n\nIn this setting we determine the minimax optimal
  learning rates (up to logarithmic factors). The rate displays interesting
  threshold behaviour: When the regression function approaches its extrema 
 at a sufficient rate\, the optimal learning rates are of the same order as
  those obtained in the label-noise free setting. If the regression functio
 n approaches its extrema more gradually then classification performance ne
 cessarily degrades. In addition\, we present an algorithm which attains th
 ese rates without prior knowledge of either the distributional parameters 
 or the local density.
LOCATION:MR12
END:VEVENT
END:VCALENDAR