A new ant colony clustering algorithm based on DBSCAN

Author

Liu, Shang ; Dou, Zhi-Tong ; Li, Fei ; Huang, Ya-lou

Author_Institution

Coll. of Inf. Sci. & Technol., Nankai Univ., Tianjin, China

Volume

3

fYear

2004

fDate

26-29 Aug. 2004

Firstpage

1491

Abstract

The AntClass algorithm is a new algorithm applying ant colony clustering algorithm to cluster analysis, and the result is satisfying. To attack the slow speed of the AntClass algorithm, a new algorithm named DBAntCluster is proposed. Firstly, the high density clusters are got in the dataset by using DBSCAN algorithm, and then these high density clusters are scattered in the grid board as a special kind of data object with other single data objects in the dataset. In DBAntCluster algorithm, the ants can avoid many unnecessary movements by using the data attribute of density and distribution well, and the speed is greatly accelerated. This improvement is validated in our experiments.

Keywords

pattern clustering; statistical analysis; DBAntCluster algorithm; DBSCAN algorithm; ant colony clustering algorithm; high density cluster analysis; Acceleration; Algorithm design and analysis; Ant colony optimization; Clustering algorithms; Educational institutions; Information processing; Information science; Laboratories; Motion analysis; Scattering;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on

Print_ISBN

0-7803-8403-2

Type

conf

DOI

10.1109/ICMLC.2004.1382009

Filename

1382009