Research And Application In Clustering Of Ant Colony Algorithm

Ant colony algorithm (ACA) is a new type bionic algorithm which simulates ants'behavior of finding foods. The algorithm has several virtues such as positive feedback and parallel mechanism, preferable robusticity, distributed computing, and easy combination with another methods. The algorithm which represents outstanding performance and tremendous developing potential attracts lots of scholars around world to study itself presently.Clustering analysis commonly called Clustering is a kind of multi-statistic analysis and is a part of important data mining. It is also an important method of data partition and grouping. The goal of clustering is to partition samples set into such clusters that intra-cluster samples are similar and inter-cluster samples are dissimilar with out any prior knowledge, which bases on samples'comparability. So clustering is also known as"unsupervised classification". Clustering has widely application such as image segmentation, medicine diagnosis, weather forecast, mineral resources identify and business affairs.In this paper, we first introduce definition, methods and data type of clustering analysis and measurement of clustering configuration. Meanwhile, we particularize several classical algorithm of every basic clustering algorithm. Then we introduce the basic ant colony algorithm and several applications of it in clustering. Finally, we have researched and improved the method of ant colony algorithm, and applied it to clustering analysis. We have improved ant colony clustering algorithm as follows: efficiency of the algorithm, quality of the algorithm, direction choice of ants'move and reducing parameter and present two ant colony clustering algorithms. Data of the experiment adopted a set of two-dimension data and IRIS to test and validate the algorithms. In order to test validity of the algorithms, they compared with K-Means and the basic ant colony clustering at the same time. At last, we conclude and analyze current research work, discuss our intending research work in this area and prospect the ant colony clustering algorithm.