An Ant Colony Clustering Algorithm Improved from ATTA

DENNEUBOURG presents the first ant-based clustering algorithm in 1991.Ant colony clustering has the characteristics of automatically clustering, high clustering accuracy, irregular cluster shapes and so on. These are important for a clustering algorithm, so ant colony clustering is arousing more and more data mining researchers’ concentration. In 2004, J. Handl together with his partners normalized the basic ant colony clustering algorithm, getting the ATTA model with superior performance. In this paper, we’ll present an algorithm improved from ATTA which we call it LCA and test LCA on UCI data sets to show its feasibility.