学习特征权值对聚类算法的优化  

Optimization of Clustering by Feature Weight Learning

作　　者：陈艳[1] 冯必波[1] 

机构地区：[1]重庆工学院计算机科学与工程学院,重庆400054

出　　处：《重庆工学院学报（自然科学版）》2008年第5期73-76,共4页Journal of Chongqing Institute of Technology

基　　金：2007年度国家火炬计划资助项目(2007GH010200)

摘　　要：现有的k-均值聚类算法大都是以距离差异为基础的,而同等重要地依赖所有属性的相似性度量会引起误导.传统的k-均值算法选择的相似性度量通常是欧几里德距离的倒数,这种距离通常涉及所有的特征.而在距离公式中引入一些特征权值后,其聚类结果将依赖于这些权值,从而可以通过调整这些权值优化聚类效果.由于k-均值算法是迭代算法,很难直接确定其权值以优化聚类结果,因此提出了一种通过免疫算法学习权值的方法以改进聚类结果.实验结果显示,该方法确定的权重值在提高聚类效果方面是可行的、有效的.The existing k-means clustering algorithms are mostly based on distance difference, but equally depending on the similarity measurement of all attributes in the same degree tends to arouse misleading. The Euclidean distance is usually chosen as the similarity measure in the conventional k-means clustering algorithm, which usually relates to all attributes. When feature weight parameters are introduced to the distance formula, the performance will depend on the weight values and accordingly can be improved by adjusting weight values. Since k-means algorithm is iterative, it is difficult to optimize clustering results by giving weight values directly. A learning feature weight algorithm based on immune algorithm is introduced to improve the clustering result. Experimental results demonstrate that this method of determining the weight values is feasible and efficient in improving the clustering performance.

关 键 词：特征权值 K-均值算法 免疫算法 聚类算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]