Hybrid immune algorithm for many optima

Abstract

In this paper an algorithm is proposed, which combines elements of immune network and clonal selection together with gradual narrowing of a search area. It is used for optimization of multi-modal functions and enables finding many optima in given domains. The algorithm introduces a novel way of interaction between memory cells and population cells. The influence of cell interaction strength on the algorithm performance has been investigated. Experiments prove that the algorithm is capable of fast localization of many optima. It outperforms other presented approaches to the multi-modal function optimization problem. © 2010 Springer-Verlag.