A fuzzy inference system to determine the number of clones in a class of artificial immune systems

Data de publicação
International Journal of Computational Intelligence and Applications
Citações (Scopus)
Carraro L.A.
De Castro L.N.
De Re A.M.
De Frana F.O.
Título da Revista
ISSN da Revista
Título de Volume
Membros da banca
Artificial immune systems are composed of techniques inspired by immunology. The clonal selection principle ensures the organism adaptation to fight invading antigens by an immune response activated by the binding of antigens and antibodies. Since the immune response must correctly allocate the available resources in order to attack an antigen with its best available antibody while trying to learning an even better one, the reproduction rate of each immune cell must be carefully determined. This paper presents a novel fuzzy inference technique to calculate the suitable number of clones for immune inspired algorithms that uses the clonal selection process as the evolutionary process. More specifically, this technique is applied to the CLONALG algorithm for solving pattern recognition tasks and to the copt-aiNet algorithm for solving combinatorial optimization tasks, particularly the Traveling Salesman Problem. The obtained results show that the fuzzy approach makes it possible to automatically determine the number of clones in CLONALG and copt-aiNet, thus eliminating this key user-defined parameter. © 2012 Imperial College Press.
Assuntos Scopus
Artificial Immune System , Clonal selection , Clonal selection principle , CLONALG , Evolutionary process , Fuzzy approach , Fuzzy inference systems , Fuzzy inference techniques , Immune cells , Immune response , Immune-inspired algorithms , User-defined parameters
DOI (Texto completo)