Union coding based immune clone selection unsupervised clustering algorithm

A new union based coding based immune clone selection clustering algorithm is proposed in this paper. The algorithm designs a coding method which synthesizes number of clusters and cluster centers, without first specifying the number of clusters, effectively overcoming the dependency on the domain knowledge. At the same time, generate clustering results by immune clone operator, and get the final results by the lopping operation of minimum spanning tree. Experimental results show that the methods can cluster for different types of data sets properly and can determine the appropriate number of clusters. © The Authors.