An improved ant colony clustering algorithm based on dynamic neighborhood

To solve the problems of the excessive clustering time consumption and the redundant numbers of the resulting clusters, commonly encountered with the ant-based clustering algorithms, an improved ant colony clustering algorithm based on dynamic neighborhood is proposed in this paper. The algorithm seeks for pure neighborhoods by performing auto-adaptive adjustments of dynamic neighborhood, and enhances ant´s memory by additionally storing the sizes of the pure neighborhoods. The ant can exchange information with other ants, load multiple similar objects at once, and merge the similar neighborhoods to form the final clusters efficiently. Experimental results indicate that this algorithm significantly improves the efficiency and quality of ant colony clustering.