基于近似分类质量及特异度的属性约简算法的部分改进  

Some Improvement of Attribute Reduction Algorithm Based on Quality of Approximation and Measure of Specificity

作　　者：安晓钢 张小红 麻卫萍 AN Xiao-gang;ZHANG Xiao-hong;MA Wei-ping(School of Arts and Sciences&Shaanxi University of Science Technology,Xi'an 710021,China;Ping An Technology(Shenzhen)Co.,Ltd.Shanghai Branch,Shanghai 200122,China)

机构地区：[1]陕西科技大学文理学院,陕西西安710021 [2]平安科技(深圳)有限公司上海分公司,上海200122

出　　处：《模糊系统与数学》2019年第5期143-151,共9页Fuzzy Systems and Mathematics

基　　金：国家自然科学基金资助项目(61573240;61473239);陕西省教育厅专项科学研究计划项目(18JK0099)

摘　　要：对基于近似分类质量及特异度的属性约简算法做了部分改进。主要解决已有算法的如下问题:当几个属性集合的近似分类质量或特异度相等、且其对应的属性组合数目也相等时,以前的算法无法分辨、只能随机选择。本文通过引入两个新的评判指标,当遇到前述问题时,计算这两个指标可以在一定程度上分别做出进一步筛选。数据结果表明,改进后算法能较快找到约简,提高了约简速度。The attribute reduction algorithm based on quality of approximation and measure of specificity is partly improved. It mainly solves the following problems of existing algorithms. When the quality of approximation or measure of specificity of several attribute sets is equal,and the number of corresponding attribute combinations is equal, the previous algorithm is unable to distinguish and can only select randomly. In this paper, two new evaluation indexes are introduced. When confronted with the aforementioned problems, the two indexes can be calculated to a certain extent to further screen the attribute sets respectively. The experimental results show that the improved algorithm can find the reduction more efficiently and speed up reduction.

关 键 词：粗糙集 近似分类质量 特异度 属性约简 

分 类 号：O159[理学—数学]