Research Area: Metaheuristic Computing
Metaheuristics are stochastic approaches to provide better solutions in a reasonable time. However, no algorithm can solve all the problems with optimality. In my thesis topic, we are interested in proposing new algorithms that can be adapted to a large number of applications. To do this, we are interested in :I- New hybridization models: because it allows to benefit from the advantages of two or more algorithms. Several hybridization models have been proposed. We try to find the best hybridization to exploit the studied algorithms.2- The self-adaptation of metaheuristic parameters: we are interested in new self-adaptation strategies that allow the algorithm to find the best parameter values that are adapted as much as possible to the problem at hand3- The integration of learning approaches: The integration of machine learning techniques into existing algorithms has become a very important area of research. Indeed, several algorithms have exploited these techniques on different contexts such as the initialization of population, the classification of the solutions and the choice between the search operators. In addition, many problems are expensive to solve. However, it is possible to shorten the computation time by using surrogate models instead of systematically using the objective function.
Name of the Researcher: Mokhtar Essaid
Name of the Supervisor(s): Lhassane Idoumghar, Julien Lepagnot
Year of Completion: 2019
University: University of Haute Alsace
Thesis Link: Home Page Url
