BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:CANCELLED: Learning Components for Human Sensing - Dr Fernando de 
 la Torre (CMU)
DTSTART:20100119T110000Z
DTEND:20100119T120000Z
UID:TALK22421@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:Providing computers with the ability to understand human behav
 ior from sensory data (e.g. video\, audio\, or wearable sensors) is an ess
 ential part of many applications that can benefit society such as clinical
  diagnosis\, human computer interaction\, and social robotics. A critical 
 element in the design of any behavioral sensing system is to find a good r
 epresentation of the data for encoding\, segmenting\, classifying and pred
 icting subtle human behavior. In this talk I will propose several extensio
 ns of Component Analysis (CA) techniques (e.g. kernel principal component 
 analysis\, support vector machines\, and spectral clustering) that are abl
 e to learn spatio-temporal representations or components useful in many hu
 man sensing tasks.\n \nIn the first part of the talk I will give an overvi
 ew of several ongoing projects in the CMU Human Sensing Laboratory\, inclu
 ding our current work on depression assessment and deception detection fro
 m video\, as well as hot-flash detection from wearable sensors. In the sec
 ond part of the talk I will show how several extensions of the CA methods 
 outperform state-of-the-art algorithms in problems such as temporal alignm
 ent of human behavior\, temporal segmentation/clustering of human activiti
 es\, joint segmentation and classification of human behavior\, and facial 
 feature detection in images.  The talk will be adaptive\, and I will discu
 ss the topics of major interest to the audience.\n 
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR