An artificial immune cell model based C-means clustering algorithm

For the problem that the classical clustering algorithm is usually sensitive to initial value or easy to bring about local optima, a novel clustering algorithm is provided which is based on models of C-means and the artificial immune mechanisms. Namely, on one hand, the process or principles that immune cells change into mature cells, and then polarize into antibodies or memory cells; on the other hand, the way or methods that an antibody captures an antigen based on the immune cellpsilas learning and remembering capabilities. Simulations show that the method proposed outperforms the classical clustering algorithm in ability of global convergence, and it appears several features such as high accuracy of clustering and better clustering capability.