Foundations of Immunocomputing

This paper presents the mathematical basis of the immunocomputing using feature extraction and pattern recognition. The key notions of the approach are the formal immune network (FIN) and the coding theory for machine learning. The training of FIN includes apoptosis (programmed cell death) and autoimmunization both controlled by cytokines (messenger proteins), whereas parameters of FIN can be optimized by Kullback entropy. Recent results suggest that the approach outperforms (by training time and accuracy) state-of-art approaches of computational intelligence.