BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Probabilistic Graphical Models for Aggregating Crowdsourced Data
DTSTART:20140616T090000
DTSTAMP:20260407T144139Z
UID:a3ac72614a6f3e2f328c2104fc83e3972e3abafa7caac6b9d5c121e0
CATEGORIES:Conferences - Seminars
DESCRIPTION:Dit-Yan YEUNG\, The Hong Kong University of Science and Techno
 logy\, Hong Kong\nCrowdsourcing is a problem-solving approach which takes 
 advantage of the wisdom of the crowd by enlisting a crowd of contributors 
 to get a task done.  Some major challenges in crowdsourcing include how t
 o evaluate the contributors and their contributions in the absence of grou
 nd-truth information and how to aggregate the contributions of multiple co
 ntributors with the goal of outperforming any single contributor.  We con
 sider two novel crowdsourcing tasks in this talk.  The first one is a vid
 eo annotation task which seeks to track an object of interest as it moves 
 around in a video.  Unlike the relatively simple classification and regre
 ssion tasks considered by most existing crowdsourcing methods\, video anno
 tation is significantly more complex because it involves structured time s
 eries data.  The second task is peer grading which is especially crucial 
 to the grading of open-ended assignments in massive open online courses (M
 OOCs).  This task is unique in that the peer graders (contributors) are t
 hemselves students whose submissions are graded by their peers.  We propo
 se probabilistic graphical models for these two crowdsourcing tasks.  Whi
 le the models are different due to the nature of the tasks\, a subtle simi
 larity is that learning the reliability of each contributor from data play
 s a central role in both machine learning models.
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR