BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Structured Prediction Models for High-level Computer Vision Tasks 
 - Sebastian Nowozin\, Microsoft Research
DTSTART:20110928T140000Z
DTEND:20110928T150000Z
UID:TALK33034@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Rich statistical models have revolutionized computer vision re
 search:\n\ngraphical models and structured prediction in particular are no
 w commonly used tools to address hard computer vision problems.  I discuss
  what distinguishes\nthese computer vision problems from other machine lea
 rning problems and how this poses unique challenges.\n\nI argue that most 
 computer vision models are misspecified and discuss the consequences of po
 pular estimators in this case\, concluding that we either\nhave to use non
 -parametric models or use estimators robust to misspecification.\n\nAs one
  possible solution\, I propose a novel discrete random field model applica
 ble to a large number of computer vision tasks.  The model is conditionall
 y specified\, non-parametric\, and able to represent complex label interac
 tions\, yet it can be trained from hundreds of images in minutes on a sing
 le machine.\n
LOCATION:Large lecture theatre\, Microsoft Research Ltd\, 7 J J Thomson Av
 enue (Off Madingley Road)\, Cambridge
END:VEVENT
END:VCALENDAR