An ACO-based Approach to Improve C-means Clustering Algorithm
Wenliang Huang, Jin Gou, Huifeng Wu · 2006
This paper presents an improved C-means clustering algorithm based on ACO. The proposed method use pheromone to evaluate individual colony's iterative result. In contrast with the existing C-means clustering algorithm, method in the paper need not appoint the number and pre-centers of clusters beforehand and it updates pheromone according to the transfer process of data points among different temporary clusters so as to avoid the local optima and reduce the iterative times to find actual cluster centers. We test its convergence performance with CRM data sets from China Unicom Corp. The experimental results show feasibility of design rationale.