SHAMS Parham
Supervision : Aurélie BEYNIER, Nicolas MAUDET
Co-supervision : BOUVERET Sylvain
Procedures based on Exchanges and new Relaxations of Envy-Freeness in Fair Division of Indivisible Goods
The work of this thesis is in the scope of Computational Social Choice. It is a field at the intersection of Social Choice, Computer Science and Artificial Intelligence. In particular, we study the problem of Fair Division of Indivisible Goods where the objective is to find a fair and efficient allocation of a set of (valuable) objects among a set of agents. While efficiency is usually brought by the minimal requirement of completeness (all the objects have to be allocated in order not to waste anything), or the more demanding notion of Pareto-Optimality (an allocation is Pareto-Optimal if there is no allocation such that all the agents are not worse off and one agent is strictly better off), several notions have been proposed to define the fairness of an allocation. One of the most prominent fairness measures is called envy-freeness. An allocation is said to be envy-free if no agent would like to exchange her bundle of resources with another agent. However, envy-freeness is not guaranteed to exist when considering indivisible goods so various relaxation has been proposed recently in the literature to overcome this limitation.
In this thesis, we first thoroughly study a family of decentralized allocation procedures related to exchanges of goods. We analyze how these procedures behave and the desirable properties they exhibit. More specifically, we study sequence of sincere choices and cycle exchanges of resources. We then propose new relaxation of the envy-freeness notion (and also of other fairness measures) and thoroughly study them. Our first relaxation aims at balancing the envy among the agents (when it cannot be avoided) and is based on the Order Weighted Average (OWA) aggregator usually used in multi-criterion optimization to bring fairness. The second relaxation focuses on the social approval of the envy and is more related to voting theory, as it lets agents vote about the envy of the other agents. We investigate computational issues related to these new relaxation, their link with existing fairness and efficiency notions and we experimentally test them.
Defence : 12/21/2023
Jury members :
Gauthier Picard, Directeur de Recherche, ONERA, Université de Toulouse [Rapporteur]
Agnieszka Rusinowska, Directrice de Recherche, Centre d'Economie de la Sorbonne, Université Paris 1 Panthéon Sorbonne [Rapporteur]
Bruno Escoffier, Professeur, LIP6, Sorbonne Université
Laurent Gourves, Directeur de Recherche, LAMSADE, Université Paris Dauphine
Aurélie Beynier, Maître de conférences HDR, LIP6, Sorbonne Université
Sylvain Bouveret, Maître de conférences, LIG, Université Grenoble-Alpes
Nicolas Maudet, Professeur, LIP6, Sorbonne Université
2019-2023 Publications
-
2023
- P. Shams : “Procedures based on Exchanges and new Relaxations of Envy-Freeness in Fair Division of Indivisible Goods”, thesis, phd defence 12/21/2023, supervision Beynier, Aurélie Maudet, Nicolas, co-supervision : Bouveret, Sylvain (2023)
-
2022
- P. Shams, A. Beynier, S. Bouveret, N. Maudet : “Fair in the Eyes of Others”, Journal of Artificial Intelligence Research, vol. 75, pp. 913-951, (Association for the Advancement of Artificial Intelligence) (2022)
-
2021
- P. Shams, A. Beynier, S. Bouveret, N. Maudet : “Minimizing and balancing envy among agents using Ordered Weighted Average”, 7th International Conference on Algorithmic Decision Theory, Toulouse, France (2021)
- A. Beynier, N. Maudet, S. Rey, P. Shams : “Swap Dynamics in Single-Peaked Housing Markets”, Autonomous Agents and Multi-Agent Systems, vol. 35 (2), pp. 20, (Springer Verlag) (2021)
-
2020
- P. Shams, A. Beynier, S. Bouveret, N. Maudet : “Fair in the Eyes of Others”, 24th European Conference on Artificial Intelligence, Santiago de Compostela, Spain (2020)
-
2019
- P. Shams, A. Beynier, S. Bouveret, N. Maudet : “Minimizing and balancing envy among agents using Ordered Weighted Average”, Actes JAIAF Journées de l’Intelligence Artificielle Fondamentale 2019, Toulouse, France, pp. 175-182, (Association française pour l'Intelligence Artificielle (AFIA)) (2019)
- A. Beynier, S. Bouveret, M. Lemaître, N. Maudet, S. Rey, P. Shams : “Efficiency, Sequenceability and Deal-Optimality in Fair Division of Indivisible Goods”, Conférence Nationale en Intelligence Artificielle, Toulouse, France (2019)
- A. Beynier, S. Bouveret, M. Lemaître, N. Maudet, S. Rey, P. Shams : “Efficiency, Sequenceability and Deal-Optimality in Fair Division of Indivisible Goods”, 18th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2019), Montreal, Canada (2019)