A New Density-Based Scheme for Clustering Based on Genetic Algorithm

Article type: Research Article

Authors: Lin, Chih-Yang | Chang, Chin-Chen | Lin, Chia-Chen

Affiliations: Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi, Taiwan 621, R.O.C. E-mail: gary,[email protected] | Department of Computer Science and Information ManagementLondon, Providence University, Taichung, Taiwan 433, R.O.C. E-mail: [email protected]

Abstract: Density-based clustering can identify arbitrary data shapes and noises. Achieving good clustering performance necessitates regulating the appropriate parameters in the density-based clustering. To select suitable parameters successfully, this study proposes an interactive idea called GADAC to choose suitable parameters and accept the diverse radii for clustering. Adopting the diverse radii is the original idea employed to the density-based clustering, where the radii can be adjusted by the genetic algorithmto cover the clusters more accurately. Experimental results demonstrate that the noise and all clusters in any data shapes can be identified precisely in the proposed scheme. Additionally, the shape covering in the proposed scheme is more accurate than that in DBSCAN.

Keywords: Clustering algorithms, genetic algorithms, DBSCAN

Journal: Fundamenta Informaticae, vol. 68, no. 4, pp. 315-331, 2005