HDR defense of Raca Todosijevic (IT department)

Rapporteurs:

• Leandro Coelho, Full professor, University of Laval, Canada.
• Ivana Ljubic, Full Professor, ESSEC Business School, France.
• Rafael Marti, Full Professor, University of Valencia, Spain.

Examiners:

• Abdessamad Ait El Cadi, Full Professor, INSA Hauts-de-France & Polytechnic University of Hauts-de-France, France.
• Jack Brimberg, Full Professor, Royal Military College of Canada, Canada.
• Bilel Derbel, Full Professor, University of Lille, France.
• Francisco Saldanha da Gama, Full Professor, Sheffield University, United Kingdom.

Research referent:

• Abdelhakim Artiba, Full Professor, Polytechnic University of Hauts-de-France, France.

This thesis offers a comprehensive review of the author's contributions to heuristic optimization, focusing on the paradigms of change and simplicity.

The concept of change, central to Heraclitus' philosophy, is a fundamental principle in the development of metaheuristics, notably Variable Neighborhood Search (VNS).

Faced with the growing complexity of hybrid heuristics, the author defends a "less-is-more" (LIMA) methodology that prioritizes simplicity and efficiency, leading to the development of powerful yet simple algorithms.

The thesis explores successful applications of VNS and LIMA to various optimization problems, highlighting the effectiveness of VNS variants (standard, Generalized, Skewed Generalized).

Each application is detailed, covering problem definition, solution approach, results and contributions.

Finally, the thesis proposes avenues for future research, including optimization under uncertainty, parallelization, matheuristics, decomposition approaches and machine learning-based heuristics, areas where the author has already made significant advances.