基于粗糙集的聚类算法中阈值自动选取

Journal of Systems Engineering and Electronics ›› 2010, Vol. 32 ›› Issue (1): 192-194.

基于粗糙集的聚类算法中阈值自动选取

宋晓宇，刘锋，孙焕良

（沈阳建筑大学信息与控制工程学院，辽宁沈阳 110168）

出版日期:2010-01-23 发布日期:2010-01-03

Autonomous threshold selection based on rough set theory in clustering algorithm

SONG Xiao-yu, LIU Feng, SUN Huan-liang

(School of Information and Control Engineering, Shenyang Jianzhu Univ., Shenyang 110168，China)

Online:2010-01-23 Published:2010-01-03

摘要/Abstract

摘要：

输入参数影响聚类算法的可用性，利用逐差法自动选取初始化阈值，使聚类算法无须任何参数，且有效降低算法的时间复杂度。逐差法利用已有数据本身属性，对相似系数矩阵行数据进行快速排序，逐个做差,求取初始化阈值。试验结果表明，新方法保证了分类精度，提高了运行效率。逐差法的应用使得基于粗糙集的聚类算法成为一种更加客观、准确的聚类方法。

Abstract:

Parameters decrease the usability of clustering algorithms. The initial threshold is selected by using a method of graded datum subtraction, so that the clustering algorithm does not need any parameter and the time complexity can be lowered. Using the attributes of known data themselves, the proposed method sorts every row of the similarity matrix quickly and makes subtraction one by one so as to acquire the initial threshold. Experiment results illustrate that the new method increases the accuracy and efficiency. The clustering algorithm based on rough set theory becomes more accurate and objective through the application of new method of graded datum subtraction.

宋晓宇，刘锋，孙焕良. 基于粗糙集的聚类算法中阈值自动选取[J]. Journal of Systems Engineering and Electronics, 2010, 32(1): 192-194.

SONG Xiao-yu, LIU Feng, SUN Huan-liang. Autonomous threshold selection based on rough set theory in clustering algorithm[J]. Journal of Systems Engineering and Electronics, 2010, 32(1): 192-194.