BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Extreme Classification: A New Paradigm for Ranking & Recommendatio
 n - Manik Varma (Microsoft Research India)
DTSTART:20150702T100000Z
DTEND:20150702T110000Z
UID:TALK59638@talks.cam.ac.uk
CONTACT:Dr Jes Frellsen
DESCRIPTION:*Abstract*\n\nThe objective in extreme multi-label classificat
 ion is to learn a classifier that can automatically tag a data point with 
 the most relevant subset of labels from a large label set. Extreme multi-l
 abel classification is an important research problem since not only does i
 t enable the tackling of applications with many labels but it also allows 
 the reformulation of ranking and recommendation problems with certain adva
 ntages over existing formulations.\n\nOur objective\, in this talk\, is to
  develop an extreme multi-label classifier that is faster to train and mor
 e accurate at prediction than the state-of-the-art Multi-label Random Fore
 st (MLRF) algorithm [Agrawal et al. WWW 13] and the Label Partitioning for
  Sub-linear Ranking (LPSR) algorithm [Weston et al. ICML 13]. MLRF and LPS
 R learn a hierarchy to deal with the large number of labels but optimize t
 ask independent measures\, such as the Gini index or clustering error\, in
  order to learn the hierarchy. Our proposed FastXML algorithm achieves sig
 nificantly higher accuracies by directly optimizing an nDCG based ranking 
 loss function. We also develop an alternating minimization algorithm for e
 fficiently optimizing the proposed formulation. Experiments reveal that Fa
 stXML can be trained on problems with more than a million labels on a stan
 dard desktop in eight hours using a single core and in an hour using multi
 ple cores.\n\n*Brief Bio*\n\nManik Varma is a researcher at Microsoft Rese
 arch India. Manik received a bachelor's degree in Physics from St. Stephen
 's College\, University of Delhi in 1997 and another one in Computation fr
 om the University of Oxford in 2000 on a Rhodes Scholarship. He then staye
 d on at Oxford on a University Scholarship and obtained a DPhil in Enginee
 ring in 2004. Before joining Microsoft Research\, he was a Post-Doctoral F
 ellow at the Mathematical Sciences Research Institute Berkeley. He has bee
 n an Adjunct Professor at the Indian Institute of Technology (IIT) Delhi i
 n the Computer Science and Engineering Department since 2009 and jointly i
 n the School of Information Technology since 2011. His  research interests
  lie in the areas of machine learning\, computational advertising and comp
 uter vision. He has served as an Area Chair for machine learning and compu
 ter vision conferences such as CVPR\, ICCV\, ICML and NIPS. He has been aw
 arded the Microsoft Gold Star award and has won the PASCAL VOC Object Dete
 ction Challenge.
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR