Supervision : Patrice PERNY, Olivier SPANJAARD

Élicitation incrémentale des préférences pour l'optimisation multi-objectifs : modèles non-linéaires, domaines combinatoires et approches tolérantes aux erreurs

This thesis work falls within the area of algorithmic decision theory, a research domain at the crossroad of decision theory, operations research and artificial intelligence. The aim is to produce interactive optimization methods based on incremental preference elicitation in decision problems involving several criteria, opinions of agents or scenarios. Preferences are represented by general decision models whose parameters must be adapted to each decision problem and each decision maker.
Our methods interleave the elicitation of parameters and the exploration of the solution space in order to determine the optimal choice for the decision maker. The idea behind this is to use information provided by the elicitation to guide the exploration of the solution space and vice versa. In this thesis, we introduce new incremental elicitation methods for decision making in different contexts$!!$: first for decision making in combinatorial domains when the decision models are non-linear, and then in a setting where one takes into account the possibility of inconsistencies in the answers of te decision maker. All the algorithms that we introduce are general and can be applied to a wide range of multiobjective decision problems.

2017-2020 Publications

• 2020
  • N. Bourdache : “Élicitation incrémentale des préférences pour l’optimisation multi-objectifs : modèles non-linéaires, domaines combinatoires et approches tolérantes aux erreurs”, thesis, defence 12/16/2020, supervision Perny, olivier spanjaard, Patrice (2020)
  • N. Bourdache, P. Perny, O. Spanjaard : “Bayesian preference elicitation for multiobjective combinatorial optimization”, DA2PL 2020 - From Multiple Criteria Decision Aid to Preference Learning, Trento, Italy (2020)
  • N. Bourdache, P. Perny, O. Spanjaard : “Élicitation incrémentale de préférences par mise à jour Bayésienne sur des zones d’optimalité”, ROADEF 2020 - 21^e congrès annuel de la société Française de Recherche Opérationnelle et d'Aide à la Décision, Montpellier, France (2020)
• 2019
  • N. Bourdache, P. Perny, O. Spanjaard : “Incremental Elicitation of Rank-Dependent Aggregation Functions based on Bayesian Linear Regression”, Proceedings of IJCAI 2019, Macao, China, pp. 2023-2029, (International Joint Conferences on Artificial Intelligence Organization) (2019)
  • N. Bourdache, P. Perny : “Algorithmes d’élicitation incrémentale des préférences pour la résolution de problèmes de sac-à-dos multi-agents équitables”, ROADEF, Le Havre, France (2019)
  • N. Bourdache, P. Perny : “Active Preference Learning based on Generalized Gini Functions: Application to the Multiagent Knapsack Problem”, Thirty-Third AAAI Conference on Artificial Intelligence (AAAI 2019), Honolulu, United States (2019)
  • N. Bourdache, P. Perny, O. Spanjaard : “Active Preference Elicitation by Bayesian Updating on Optimality Polyhedra”, SUM 2019 - 13^th international conference on Scalable Uncertainty Management, vol. 11940, Lecture Notes in Computer Science, Compiègne, France, pp. 93-106, (Springer) (2019)
• 2018
  • N. Bourdache, P. Perny, O. Spanjaard : “Adaptive Elicitation of Rank-Dependent Aggregation Models based on Bayesian Linear Regression”, DA2PL'2018, Poznan, Poland (2018)
• 2017
  • N. Bourdache, P. Perny : “Anytime Algorithms for Adaptive Robust Optimization with OWA and WOWA”, 5^th International Conference on Algorithmic Decision Theory (ADT 2017), vol. 10576, Lecture Notes in Computer Science, Luxembourg, Luxembourg, pp. 93-107, (Springer) (2017)