An improved self-organization antibody network for pattern recognition and its performance study

Abstract

Self-organization antibody network (soAbNet) is a novel artificial immune system inspired by the biological immune system. This paper presents an modified soAbNet for pattern classification. The work here builds upon previous work on artificial immune network (AIS) for pattern reorganization. A population control mechanism has been introduced to optimize immune network architecture similarly to that in resource limited artificial immune system (RLAIS). Contrast experiments on Iris dataset are performed to analyze the performance of soAbNet. Experimental results demonstrate the proposed algorithm efficiently compresses network scale with high classification accuracy at the same time. © 2012 Springer-Verlag.