Till startsida
To content Read more about how we use cookies on gu.se

Angeliki Lazaridou "Emergent communication through multi-agent interactions"

Research profile seminar

Angeliki Lazaridou "Emergent communication through multi-agent interactions"

Supervised models of language focus on the structure of language and are an excellent way to learn general statistical associations between sequences of symbols. However, they do not capture the functional aspects of communication, i.e., that humans have intentions and use words to coordinate with others and make things happen in the real world.

In this talk, I will present two studies on multi-agent emergent communication through reinforcement learning. Agents exist in a grounded environment and have to achieve goals, a process that requires them to communicate about their beliefs in the pragmatic context of achieving these goals. In the first study, I will present experiments in which we analyze the communication conventions and their properties formed by agents that learn to communicate about symbolic (i.e., attribute-based) and entangled (i.e., raw pixel) input. In the second study, I will talk about properties of linguistic communication as arising in the context of negotiating self-interested and prosocial agents.

Lecturer: Angeliki Lazaridou

Date: 3/14/2018

Time: 1:15 PM - 3:00 PM

Categories: Linguistics

Organizer: CLASP

Location: T307, Olof Wijksgatan 6

Contact person: stergios chatzikyriakidis

Page Manager: |Last update: 9/15/2015

The University of Gothenburg uses cookies to provide you with the best possible user experience. By continuing on this website, you approve of our use of cookies.  What are cookies?