摘要
传统的聚类算法存在很多缺点，因此需要做进一步的研究。通过对传统的K-means算法和加权熵措施的K—means算法的研究，提出了一种改进的加权熵措施的K—means算法，且该算法采用了一种新的计算对象间距离的方法，不仅能使在同一个簇中任意对象之间的距离尽可能的小，更能使得不同簇中的任意对象之间的距离尽可能的大。通过在KDD Cup99数据集上实验仿真，表明该算法具有较强的实用性和自适应功能。
Traditional clustering algorithm has a lot of shortcomings ,therefore need to do further study.Through studying the traditional K-means algorithm and the K-means algorithm of entropy-weighted measure, an improved K-means algorithm of entropy-weighted measure is proposed, the algorithm uses a new method of calculating the distance of the objects not only make the distance between any objects close as much as possible in the same cluster, but also make the distance between any objects as large as possible in the different clusters. Through the KDD Cup99 data set simulation experiment, showing that the algorithm has a strong applicability and sell-adaptability.
作者
王彦涛
张凤斌
WANG Yan-tao,ZHANG Feng-bin （School of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, China）
出处
《电脑知识与技术》
2009年第12期9824-9827,共4页
Computer Knowledge and Technology
基金
国家自然科学基金项目（60671049）
