BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian modeling for high-level real nursing activity recognition
  using accelerometers - Prof. Naonori Ueda (Director Machine Learning and 
 Data Science\, NTT Labs)
DTSTART:20141114T110000Z
DTEND:20141114T113000Z
UID:TALK56184@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:When we face new complex classification tasks\, since it is di
 fficult to design a good feature set for observed raw data\, we often obta
 in an unsatisfactorily\nbiased classifier. Namely\, the trained classifier
  can only successfully classify certain classes of samples owing to its po
 or feature set. To tackle the problem\, we propose a robust naive Bayes co
 mbination scheme in which we effectively combine classifier predictions th
 at we obtained from different classifiers and/or different feature sets. S
 ince we assume that the multiple classifier predictions are given\, any ty
 pe of classifier and any feature set are available in our scheme. In our c
 ombination scheme each prediction is regarded as an independent realizatio
 n of a  categorical random variable (i.e.\, class label) and a naive Bayes
  model is trained by using a set of the predictions within a supervised le
 arning framework. The key feature of our scheme is the introduction\nof a 
 class-specific variable selection mechanism to avoid overfitting to poor c
 lassifier predictions. We demonstrate the practical benefit of our simple 
 combination scheme with both synthetic and real data sets\, and show that 
 it can achieve much higher classification accuracy than conventional ensem
 ble classifiers. We apply this method to high-level real nursing activity 
 recognition in a hospital using accelerometers\, and show the usefuleness 
 of the method.
LOCATION:Engineering Department\, CBL Room BE-438.
END:VEVENT
END:VCALENDAR