BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Emergence of (linguistic) communication through multi-agent intera
 ctions - Angeliki Lazaridou\, DeepMind
DTSTART:20181122T110000Z
DTEND:20181122T120000Z
UID:TALK114934@talks.cam.ac.uk
CONTACT:Edoardo Maria Ponti
DESCRIPTION:*Abstract*: Distributional models and other supervised models 
 of language focus on the structure of language and are an excellent way to
  learn general statistical associations between sequences of symbols. Howe
 ver\, they do not capture the functional aspects of communication\, i.e.\,
  that humans have intentions and use words to coordinate with others and m
 ake things happen in the real world. In this talk\, I will present two stu
 dies on multi-agent emergent communication\, where agents exist in some gr
 ounded environment and have to communicate about objects and their propert
 ies. This process requires the negotiation of linguistic meaning in this p
 ragmatic context of achieving their goal. In the first study\, I will pres
 ent experiments in which agents learn to form a common ground that allow t
 hem to communicate about disentangled (i.e.\, feature norm) and entangled 
 (i.e.\, raw pixels) input. In the second study\, I will talk about propert
 ies of linguistic communication as arising in the context of self-interest
 ed agents.
LOCATION:Boardroom\, Faculty of English\, West Road
END:VEVENT
END:VCALENDAR