Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19]


Journal of Information Science and Engineering, Vol. 24 No. 5, pp. 1445-1460 (September 2008)

A Two-stage Clustering Method Combining Ant Colony SOM and K-means*

Sheng-Chai Chi and Chih-Chieh Yang+
Department of Industrial Engineering and Management Information
Huafan University
Taipei County, 223 Taiwan
+Advantech Corp.
Taipei, 114 Taiwan

Self-organizing map (SOM), which is an orderly mapping technique, can convert complex, non-linear and high-dimensional data into simple, geometric and low-dimen- sional data that can easily be visualized. In data analysis techniques, the capability of SOM and K-means for clustering large-scale databases has already been confirmed. Although SOM and K-means have their superior features for cluster analysis, their combination into a two-stage method is generally much more powerful than the two methods used individually. In this research, an ant-based self-organizing map (ABSOM) is proposed. The ABSOM embeds the exploitation and exploration rules of state transition into the conventional SOM algorithm to avoid falling into local minima. To examine the usefulness of the proposed method, the ABSOM is combined with K-means into a two- stage clustering method, i.e. ABSOM+K-means. Applied four public data sets, the ABSOM has been proved that it performs better than KohoneníŽs SOM and it also works very well in the two-stage cluster analysis when it is taken as a preprocessing technique.

Keywords: SOM, ant colony system, K-means, U-matrix, clustering, two-stage method

Full Text (ą■Ąň└╔) Retrieve PDF document (200809_10.pdf)

Received December 5, 2006; revised April 24 & June 25, 2007; accepted August 3, 2007.
Communicated by Pau-Choo Chung.
*The preliminary version has appeared in the Proceedings of the 10th International Conference on Knowledge-Based Intelligent Information & Engineering Systems, KESíŽ2006. The work was partially supported by the National Science Council of Taiwan, R.O.C. under grant No. NSC 94-2213-E-211-009.