Mhc-Inspired Antibody Clone Algorithm
YU ZHANG,XUMING CHEN,LIHUA WU,ZIQIANG LUO,XIAOJIE LIU
DOI: https://doi.org/10.1142/S0219876210002179
IF: 1.7
2010-01-01
International Journal of Computational Methods
Abstract:For solving complex optimization problems in some engineering applications, intelligent optimization algorithms based on biological mechanisms have better performance than traditional optimization algorithms. Most of these intelligent algorithms, however, have disadvantages in population diversity and preservation of elitist antibody genes, which lead to the degenerative phenomenon, the zigzag phenomenon, poor global optimization, and low convergence speed. Drawing inspiration from the features of major histocompatibility complex (MHC) in the biological immune system, we propose a novel MHC-inspired antibody clone algorithm (ACAMHC) for solving the above problems. ACAMHC preserves elitist antibody genes through the MHC strings that emulate the MHC haplotype in order to improve its local search capability; it improves the antibody population diversity by gene mutation that mimick the MHC polymorphism to enhance its global search capability. To expand the antibody search space, ACAMHC will add some new random immigrant antibodies with a certain ratio. The convergence of ACAMHC is theoretically proven. The experiments of ACAMHC compared with the canonical clone selection algorithm (CLONALG) on 20 benchmark functions are carried out. The experimental results indicate that the performance of ACAMHC is better than that of CLONALG. The ACAMHC algorithm provides new opportunities for solving previously intractable optimization problems.