Current Topics of Interest
Computational Social Choice— At the interface of computer science (A.I., theoretical computer science, logic) and economics (social choice theory), this relatively recent field studies various aspects of collective decision problems, where several agents have preferences over possible outcomes, or candidates. This includes in particular voting (chosing a single candidate), and resource allocation (sharing resources among agents)–but also many more things. My interest in this field lies more specifically in distributed mechanisms for fair division of resources, specific voting settings where candidates may be strategic themselves, and communication issues in general.
- ANR project CoCoRiCoCODEC ``Calcul, Communication, Rationalité et Incitations en Décision Collective et Coopérative’’ (start: Octobre 2014)
- COST action on Computational Social Choice (COMSOC, ended in 2016)
Models of Collective Argumentation and Deliberation— Either individually or as the result of a collective debate (as provided for instance by online deliberation platforms), it is common to have to handle sets of (typically conflicting) arguments. Argumentation theory provides the fundamentals to handle such argumentation networks and offers many alternative approaches, but the multiagent aspect remains challenging in many ways. How is it possible to make sense of such argumentation networks? Can we identify strong or important arguments? Are there any sensible way to regulate such debates? Can we identify malicious users?
- ANR project on Advanced Multilateral Argumentation for DEliberation (AMANDE, start: December 2013);
- PhD thesis of Jérôme Delobelle (defense planned: Fall 2017)
Accountability and Explanation of Algorithmic Decisions— While (semi-) automated decisions (or recommendations) have become ubiquitous nowadays, there is a legitimate demand to equip such systems with explanatory capabilities. There are many difficulties with this though. First, the very notion of explanation may greatly vary depending on the user considered. On top of that, the underlying model might be complex to grasp. Deciding what and how to present to the user is a difficult question.
- PhD thesis of Khaled Belahcene (defense planned: Fall 2018)
- PhD thesis of Karim El Mernissi (defense planned: Winter 2017)
- Associate Editor of the Journal of Autonomous Agents and Multi-Agent Systems (JAAMAS)
- Editorial Board member of the Journal of Artificial Intelligence Research (JAIR)
Program Committee Membership
I am also regularly involved in the peer-review process of A.I. conferences, in particular:
- International Joint Conference of Artificial Intelligence [2009, 2011 (SPC), 2013, 2014, 2016, 2017 (SPC)]
- International Conference on Autonomous Agents and Multiagent Systems [2005, 2006 (SPC), 2007, 2008, 2009 (best PC member award), 2010, 2011, 2012, 2013, 2014, 2015, 2016 (SPC), 2017 (SPC)]
- AAAI Conference on Artificial Intelligence [2008, 2010, 2011 (SPC), 2013, 2015, 2016, 2017, 2018]