Contributions à la Résolution de l'Emergence Inversée en Utilisant les Métaheuristiques Quantiques

Abstract

Complex systems present an interesting property : emergence. This property exists in several models inspired from nature. Within these models : cellular automata, ant colonies, multi-agent systems, particle swarms, arti_cial neural networks. The resolution of reverse emergence in these systems consists in extracting the basic rules which allowed the collectif of simple individuals to cooperate and produce at a global level an emergent function. In the litterature, this problem is quali_ed as "di_cult" and induces intense research. In this context, we propose in this thesis three resolution methods. The _rst method is based on PSO metaheuristic, which guides an evolutionary cellular automata. The validation is done through the extraction of edges on images. The second method uses quantum computing principles, hybridized with PSO, producing Quantum PSO metaheuristic. The purpose is to take advantage from diversi_cation of population, parallelism, and richness of quantum operators. This model is used to resolve two problems : edge detection and _ltering of images. The third method is an hybridization between genetic algorithm and quantum computing. The obtained algorithm shows a good global search ability. A reduced number of quantum chromosomes is su_cient to cover the problem. The experimental results obtained by the three methods demonstrate the good capacity of the metaheuristics used, to provide a satisfactory solution to the problem of reverse emergence, and insure excellent convergence of the system and a good balance between exploitation and exploration