An Effective Clustering Algorithm With Ant Colony

This paper proposes a new clustering algorithm based on ant colony to solve the unsupervised clustering problem. Ant colony optimization (ACO) is a population-based meta-heuristic that can be used to find approximate solutions to difficult combinatorial optimization problems. Clustering Analysis, which is an important method in data mining, classifies a set of observations into two or more mutually exclusive unknown groups. This paper presents an effective clustering algorithm with ant colony which is based on stochastic best solution kept--ESacc. The algorithm is based on Sacc algorithm that was proposed by P.S. Shelokar. It's mainly virtue that best values iteratively are kept stochastically. Moreover, the new algorithm using Jaccard index to identify the optimal cluster number. The results of several times experiments in three datasets show that the new algorithm-ESacc is less in running time, is better in clustering effect and more stable than Sacc. Experimental results validate the novel algorithm's efficiency. In addition, Three indices of clustering validity analysis are selected and used to evaluate the clustering solutions of ESacc and Sacc.