论文信息 - A Clustering Algorithm for Datasets with Different Densities

A Clustering Algorithm for Datasets with Different Densities

Cluster analysis is a hot topic of data mining, and numerous methods or algorithms have been researched. In this article, a new cluster algorithm, called DDD algorithm, based on the decomposition of Delaunay Diagrams is proposed. The DDD algorithm can get non-spherical clusters other than spherical ones, and it can get clusters of high quality even if with lots of noise data, and it can get clusters with different density. The simulation experiments illustrate the above points.

[1] Ki-Joune Li,et al. A spatial data mining method by Delaunay triangulation , 1997, GIS '97.

[2] Olga Sourina,et al. Effective clustering and boundary detection algorithm based on Delaunay triangulation , 2008, Pattern Recognit. Lett..

[3] Nicolai Petkov,et al. Algorithm That Mimics Human Perceptual Grouping of Dot Patterns , 2005, BVAI.

[4] Ickjai Lee,et al. AMOEBA: HIERARCHICAL CLUSTERING BASED ON SPATIAL PROXIMITY USING DELAUNATY DIAGRAM , 2000 .

[5] Marina L. Gavrilova,et al. CRYSTAL - A new density-based fast and efficient clustering algorithm , 2006, 2006 3rd International Symposium on Voronoi Diagrams in Science and Engineering.

[6] Craig Eldershaw,et al. Cluster Analysis using Triangulation , 1997 .

[7] Ickjai Lee,et al. AUTOCLUST: Automatic Clustering via Boundary Extraction for Mining Massive Point-Data Sets , 2000 .