Preference Modelling in Multiagent Systems

Prof. Kate Larson - U of Waterloo

Nov. 30, 2018, 2:30 p.m. - Nov. 30, 2018, 3:30 p.m.

Trottier 2100

Hosted by: Prof. Joelle Pineau

Speaker Bio: Kate Larson is a Professor at the Cheriton School of Computer Science University of Waterloo. Her research interests include artificial intelligence and multiagent systems with a particular focus on algorithmic game theory group decision making and preference modelling. She holds the Pasupalak AI Fellowship at the University of Waterloo was a Cheriton Faculty Fellow (2012-2015) and was awarded the Canadian Association of Computer Science Outstanding Young Researcher Award.  She likes organizing groups of agents and so is the Past-President of the International Foundation for Agents and Multiagent Systems was the General Co-Chair of the 2017 International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2017) and previously served as an AAAI Councilor. 
Abstract: The central question in multiagent systems (MAS) is how should we design systems or platforms for interacting agents or decision makers such that the outcomes reached appropriately reflect the preferences and values of the agents. To address this question we require both strong agent models that capture individual preferences and values coupled with normative systems that aggregate information from the agents and make decisions in fair and representative ways.
In this talk I will present some of our work on issues which arise in the design of MAS when the agents due to missing information cognitive costs the context in which they find themselves etc are not able or are not willing to fully articulate their values or preferences. We show that this has repercussions both on agent-design and behaviour as well as the platforms we can support.