文章基本信息

标题：Adaptive clustering with artificial ants
本地全文：下载
作者：D. A. Ingaramo ; G. Leguizamón ; M. Errecalde 等
期刊名称：Journal of Computer Science and Technology
印刷版ISSN：1666-6046
电子版ISSN：1666-6038
出版年度：2005
卷号：5
期号：4
出版社：Iberoamerican Science & Technology Education Consortium
摘要：Clustering task aims at the unsupervised classi-fication of patterns (e.g., observations, data, vec-tors, etc.) in di.erent groups. Clustering problemhas been approached from di.erent disciplinesduring the last years. Although have been pro-posed di.erent alternatives to cope with cluster-ing, there also exists an interesting and novel fieldof research from which di.erent bio-inspired al-gorithms have emerged, e.g., genetic algorithmsand ant colony algorithms. In this article we pro-pose an extension of the AntTree algorithm, anexample of an algorithm recently proposed for adata mining task which is designed following theprinciple of self-assembling behavior observed insome sp ecies of real ants. The extension proposedcalled Adaptive-AntTree (AAT for short) repre-sents a more .exible version of the original one.The ants in AAT are able of changing the as-signed position in previous iterations in the treeunder construction. As a consequence, this newalgorithm builds an adaptive hierarchical clusterwhich changes over the run in order to improvethe final result. The AAT performance is experi-mentally analyzed and compared against AntTreeand K-means which is one of the more popularand referenced clustering algorithm
关键词：computational intelligence; bio-;inspired algorithms; clustering; data mining