An Improved Artificial Immune System (AIS) by Considering Different Affinities among Th Cells and Antigens

抄録

This paper introduces a new artificial immune system (AIS) model to pattern recognition. Inspired by natural immune system, we propose the AIS model, by considering not only the B cells' T-cell-dependent and T-cell-independent antigen recognition, but also the different affinities among Th cells and antigens, and apply the model to pattern recognition. We present noisy binary patterns to test its recognition performance. Simulation results indicate that the proposed AIS model has more stable memory capabilities and stronger noise tolerance ability.