A Practical Clustering Algorithm

We present a novel clustering algorithm (SDSA algorithm) based on the concept of the short distance of the consecutive points and the small angle between the consecutive vectors formed by three adjacent points. Not only the proposed SDSA algorithm is suitable for almost all test data sets used by Chung and Liu for point symmetry-based K-means algorithm (PSK algorihtm) and their newly proposed modified point symmetry-based K-means algorithm (MPSK algorithm ), the proposed SDSA algorithm is also suitable for many other cases where the PSK algorihtm and MPSK algorithm can not be well performed. Based on some test data sets, experimental results demonstrate that our proposed SDSA algorithm is rather encouraging when compared to the previous PSK algorithm and MPSK algorithm.

[1]  Ana L. N. Fred,et al.  A New Cluster Isolation Criterion Based on Dissimilarity Increments , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  Anil K. Jain,et al.  Algorithms for Clustering Data , 1988 .

[3]  Kuo-Liang Chung,et al.  An efficient line symmetry-based K-means algorithm , 2006, Pattern Recognit. Lett..

[4]  Joachim M. Buhmann,et al.  Bagging for Path-Based Clustering , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[5]  Chien-Hsing Chou,et al.  Short Papers , 2001 .

[6]  John A. Hartigan,et al.  Clustering Algorithms , 1975 .

[7]  Richard O. Duda,et al.  Pattern classification and scene analysis , 1974, A Wiley-Interscience publication.

[8]  Kuo-Liang Chung,et al.  Faster and more robust point symmetry-based K-means algorithm , 2007, Pattern Recognit..

[9]  Lai-Man Po,et al.  Minimax partial distortion competitive learning for optimal codebook design , 1998, IEEE Trans. Image Process..

[10]  Narendra Ahuja,et al.  Location- and Density-Based Hierarchical Clustering Using Similarity Analysis , 1998, IEEE Trans. Pattern Anal. Mach. Intell..