BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Towards Intelligent Agents That Can See\, Talk\, and Act 
X-WR-TIMEZONE:Pacific Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260517T213934Z
UID:tag:localist.com\,2008:EventInstance_4407451
DTSTART:20190311T170000Z
DTEND:20190311T180000Z
DESCRIPTION:Stefan Lee\, Research Scientist\nSchool of Interactive Computin
 g\nGeorgia Tech\n\nAbstract\nFor AI agents to fully step into the role of 
 human collaborators\, they must be able to perceive their environment and 
 communicate about this understanding with humans in order to coordinate th
 eir actions to achieve mutual goals. The development of such holistic agen
 ts presents challenging problems for computer vision\, natural language pr
 ocessing\, and machine learning. Towards this end\, I'll discuss a recent 
 line of work developing agents that communicate in natural language regard
 ing visual scenes including both static images and 3D environments. First\
 , I will focus on work developing agents that engage in visually-grounded\
 , question-answer based dialogs -- a task we call Visual Dialog. I will pr
 ovide an overview of the Visual Dialog task and highlight some challenges 
 faced by deep agents trained for this problem. Then I will discuss follow-
 up work in which we address some of these challenges by modeling Visual Di
 alog as a cooperative game between agents in a reinforcement learning sett
 ing -- learning dialog agent policies end-to-end\, from pixels to multi-ag
 ent\, multi-round dialog to game reward. Finally\, I'll discuss EmbodiedQA
 \, a recent effort to extend beyond static images and ground similar agent
 s into simulated 3D environments.\n\nBio\nStefan Lee is a Research Scienti
 st in the School of Interactive Computing at Georgia Tech where he studies
  problems at the intersection of machine learning\, computer vision\, and 
 natural language processing. His current work addresses how to develop age
 nts that can see\, talk\, and act -- designing agents that can understand 
 and use visually-grounded language to achieve goals in complex environment
 s. His work frequently appears at major conferences in computer vision\, n
 atural language processing\, and machine learning. He is the recipient of 
 a Best Paper award (EMNLP 2017) and was recognized as a 2018 DARPA Riser f
 or the potential impact of his research agenda. He has also received multi
 ple outstanding reviewer awards (2017 - CVPR\, ICCV\, ECCV\, NuerIPS. 2018
  - NeurIPS\, ICLR\, 2019 - ICLR) recognizing his service efforts in the co
 mmunity. Prior to his current position\, he was a Bradley Postdoctoral Fel
 low at Virginia Tech after receiving his PhD in 2016 from the School of In
 formatics and Computing at Indiana University advised by David Crandall.
GEO:44.567164;-123.278692
LOCATION:Kelley Engineering Center\, 1007
SUMMARY:Towards Intelligent Agents That Can See\, Talk\, and Act 
URL;VALUE=URI:https://events.oregonstate.edu/event/towards_intelligent_agen
 ts_that_can_see_talk_and_act
CATEGORIES:Lecture or Presentation
END:VEVENT
END:VCALENDAR