Robust classification of immunity clonal synergetic network inspired by fuzzy integral

Abstract

The selection of prototype pattern vectors and the reconstruction of order parameters are choke points in Synergetic Neural Network (SNN). We have improved the performance of SNN on these two points by Immunity Clonal Strategy (ICS) and applied them to classification successfully. But how to aggregate them remains an open question. Inspired by decision fusion mechanism, a new Immunity Clonal Synergetic Network is proposed to solve the two problems simultaneously. With the use of fuzzy integral, not only are the classification results combined but that the relative importance of the different networks is also considered. Experiments show that the presented algorithm has higher recognition rate and is more robust to classification. © Springer-Verlag Berlin Heidelberg 2005.