Exploration de nouveaux horizons en optimisation combinatoire multi-objectifs : de la génération de Pareto aux méthodes interactives
This thesis summarizes our contributions to multi-objective combinatorial optimization problems. After a first introductory chapter, the second chapter is devoted to methods (exact and heuristic) for generating the set of Pareto-optimal solutions. In the next chapter, we study a way of integrating fairness into multi-objective problems using Lorenz dominance. In the fourth chapter, we examine the use of the Choquet integral to restrict the size of the solutions set. In the final chapter, we study interactive methods, enabling interaction with the decision-maker throughout the solution generation process. Different interactive methods are presented, enabling both preference learning and solution generation. Finally, we examine different perspectives with the aim of defining future directions for research in multi-objective combinatorial optimization.
Phd defence : 03/21/2024
Jury members :
Clarisse Dhaenens, Université de Lille, [Rapporteur]Jin-Kao Hao, Université d'Angers, [Rapporteur]
Daniel Vanderpooten, Université Paris Dauphine-PSL, [Rapporteur]
Vincent Mousseau, Université Paris-Saclay, Examinateur
Luís Paquete, University of Coimbra, Examinateur
Patrice Perny, Sorbonne Université, Examinateur
Associate Professor [HDR]