BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Hedging Against Uncertainty via Multiple Diverse Predictions  - Dh
 ruv Batra\, Virginia Tech
DTSTART:20140904T130000Z
DTEND:20140904T140000Z
UID:TALK54112@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:What does a young child or a high-school student with no knowl
 edge of probability do when faced with a problem whose answer they are unc
 ertain of? They make guesses. \n\nModern machine perception algorithms (fo
 r object detection\, pose estimation\, or semantic scene understanding)\, 
 despite dealing with tremendous amounts of ambiguity\, do not. \n\nIn this
  talk\, I will describe a line of work in my lab where we have been develo
 ping machine perception models that output not just a single-best solution
 \, rather a /diverse/ set of plausible guesses. I will discuss inference i
 n graphical models\, connections to submodular maximization over a "doubly
 -exponential" space\, and how/why this achieves state-of-art performance o
 n Pascal VOC 2012 segmentation dataset. Following my own advice\, I will t
 alk about talk some other cool things as well (including deep learning of 
 course). \n
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR