BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Discriminative Embeddings of Latent Variable Models for Structured
  Data - Prof Le Song (Georgia Tech)
DTSTART:20160725T100000Z
DTEND:20160725T110000Z
UID:TALK66683@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:Kernel classifiers and regressors designed for structured data
 \, such as sequences\, trees and graphs\, have significantly advanced a nu
 mber of interdisciplinary areas such as computational biology and drug des
 ign. Typically\, kernels are designed beforehand for a data type which eit
 her exploit statistics of the structures or make use of probabilistic gene
 rative models\, and then a discriminative classifier is learned based on t
 he kernels via convex optimization. However\, such an elegant two-stage ap
 proach also limited kernel methods from scaling up to millions of data poi
 nts\, and exploiting discriminative information to learn feature represent
 ations. \n\nIn this talk\, I will present structure2vec\, an effective and
  scalable approach for structured data representation based on the idea of
  embedding latent variable models into feature spaces\, and learning such 
 feature spaces using discriminative information. Interestingly\, structure
 2vec extracts features by performing a sequence of function mappings in a 
 way similar to graphical model inference procedures\, such as the mean fie
 ld and belief propagation algorithm. In applications involving millions of
  data points\, we showed that structure2vec runs 2 times faster\, produces
  models which are 10\, 000 times smaller\, while at the same time achievin
 g the state-of-the-art predictive performance.\n\nBio: Le Song is an assis
 tant professor in the Department of Computational Science and Engineering\
 , College of Computing\, Georgia Institute of Technology. He received his 
 Ph.D. in Machine Learning from University of Sydney and NICTA in 2008\, an
 d then conducted his post-doctoral research in the Department of Machine L
 earning\, Carnegie Mellon University\, between 2008 and 2011. Before he jo
 ined Georgia Institute of Technology\, he was a research scientist at Goog
 le. His principal research direction is machine learning\, especially nonl
 inear methods and probabilistic graphical models  for large scale and comp
 lex problems\, arising from artificial intelligence\, social network analy
 sis\, healthcare analytics\, and other interdisciplinary domains. He is th
 e recipient of the NSF CAREER Award’14\, AISTATS'16 Best Student Paper A
 ward\, IPDPS'15 Best Paper Award\, NIPS’13 Outstanding Paper Award\, and
  ICML’10 Best Paper Award. He has also served as the area chair for lead
 ing machine learning conferences such as ICML\, NIPS and AISTATS\, and the
  action editor for JMLR.
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR