Mohamad EL SAYAH : "Random Generation for the Performance Evaluation of Scheduling Algorithms"

PhD Works of Moham EL SAYAH : "Random Generation for the Performance Evaluation of Scheduling Algorithms

Abstract :

The objective of the work in this theory has highlighted the techniques of performance evaluation in the context of progression. In many branches of computer research, numerous publications have been put forward for new algorithms. but the experimental protocol has been used, either little or too described, to be reproducible, or to give sufficient confidence in the extent of the improvement. This implies to be analyzed with precision.

In this thesis, we have two types of instances, DAGs and cost matrices.

For each type of examination, we have a different problem. We study the properties that tend to influence the structure of randomly generated instances and then analyze the previous generation methods with random control. Then, we propose a test of behavior pending, thus we analyze the impact of the generators studied on certain heuristic performances of scheduling.

Jury composition :

Mr Beaumont Olivier, Director of research at Inria Bordeaux - Sud-Ouest, Reviewer
Mr LEGRAND Arnaud, Director of research at CNRS Grenoble, Reviewer
Mr NICOD Jean-Marc, Professor at ENSMM Besançon, Examiner
Ms DUFOSSÉ Fanny, Research fellow at Inria Grenoble, Examiner
Mr CANON Louis-Claude, Associate professor at the University of Franche-Comté, Co-supervisor of thesis
Mr HÉAM Pierre-Cyrille, Professor at the University of Franche-Comté, Supervisor of thesis

Localization : Amphithéâtre B, Bâtiment B de l'UFR Sciences et Techniques - 16, Route de Gray - 25000 Besançon