Novel Hybrid Clustering Algorithm Incorporating Artificial Immunity into Fuzzy Kernel Clustering for Pattern Recognition

Abstract

The application of artificial immunity and fuzzy kernel clustering in data classification is studied, and a new hybrid clustering algorithm incorporating artificial immunity into fuzzy kernel clustering for pattern recognition is proposed in this paper. The algorithm, by combining kernel-based fuzzy clustering with artificial immune evolution algorithm, which learns from the mechanism of immunocyte clone, memory and affinity maturation in natural immune system, operates on antibody with clone, hyper-mutation and restraint in each generation. The algorithm can quickly obtain global optima, and perfectly solve the flaws of the fuzzy c-means and kernel clustering algorithm, which are sensitive to initialization and easy to involve local optima. Our experiments on IRIS data as well as compressor fault data demonstrate the feasibility and effectiveness of the new algorithm.