Sites Inria

Version française

DOLPHIN Research team

Parallel Cooperative Multi-criteria Optimization

Team presentation

Many industrial domains are concerned by large and complex optimization problems, involving important financial costs and in which decisions must be taken in an optimal way. The development of advanced hybrid optimization methods from combinatorial optimization in operations research, decision in artificial intelligence and parallel and distributed computing, is an important issue in solving in a reasonable search time this class of problems, which are more and more complex.

The goal of the DOLPHIN project-team is the modeling and parallel resolution of large (multi-objective) combinatorial optimization problems. Efficient parallel cooperative optimization methods are developed from the analysis of the structure of the solved problem. The target optimization problems are generic problems (flow-shop scheduling, vehicle routing, etc.) and industrial problems from logistics, transportation, energy, and bioinformatics.

Research themes

  • Analysis of the structure of a combinatorial optimization problem, where many indicators are used to analyze the landscape of the problem. This allows the efficient design of search operators, objective finctions, and hybrid methods for solving mono-objective and multi-objective problems.
  • Cooperation of optimization methods (metaheuristics and/or exact methods), to combine complementary optimization strategies.
  • Parallel optimization methods, to speedup the search, to solve large problems, and to improve the robustness and the quality of solutions.

International and industrial relations

  • EDF, GDF-Suez, Tasker, energy (electricity, cloud).
  • Genes Diffusion, Alicante, bioinformatics.
  • DHL, Vekia, Opalean, logistics and transportation.
  • Univ. Malaga (Espagne), Univ. Luxembourg, Georgia Tech (USA), Univ. Montreal (Canada), etc.
  • EvoNet European Network, Univ. Malaga (Espagne), Illinois (USA), etc.