Co-evolution-based immune clonal algorithm for clustering

Abstract

Clustering is an important tool in data mining process. Fuzzy $$c$$c-means is one of the most classic methods. But it has been criticized that it is sensitive to the initial cluster centers and is easy to fall into a local optimum. Not depending on the selection of the initial population, evolutionary algorithm is used to solve the problems existed in original fuzzy $$c$$c-means algorithm. However, evolutionary algorithm emphasizes the competition in the population. But in the real world, the evolution of biological population is not only the result of internal competition, but also the result of mutual competition and cooperation among different populations. Co-evolutionary algorithm is an emerging branch of evolutionary algorithm. It focuses on the internal competition, while on the cooperation among populations. This is more close to the process of natural biological evolution and co-evolutionary algorithm is a more excellent bionic algorithm. An immune clustering algorithm based on co-evolution is proposed in this paper. First, the clonal selection method is used to achieve the competition within population to reconstruct each population. The internal evolution of each population is completed during this process. Second, co-evolution operation is conducted to realize the information exchange among populations. Finally, the iteration results are compared with the global best individuals, with a strategy called elitist preservation, to find out the individual with a highest fitness value, that is, the result of clustering. Compared with four state-of-art algorithms, the experimental results indicate that the proposed algorithm outperforms other algorithms on the test data in the highest accuracy and average accuracy.