基于邻域关系的知识粒度增量式属性约简算法  

Knowledge granularity incremental attribute reduction algorithm based on neighborhood relation

作　　者：陈曦[1,2] 刘晶[2] CHEN Xi;LIU Jing(School of Management,Huazhong University of Science and Technology,Wuhan 430074,China;Wuhan Railway Vocational and Technical College,Wuhan 430200,China)

机构地区：[1]华中科技大学管理学院,湖北武汉430074 [2]武汉铁路职业技术学院,湖北武汉430200

出　　处：《微电子学与计算机》2020年第10期1-6,共6页Microelectronics & Computer

基　　金：国家自然科学基金(71277097)。

摘　　要：为了在邻域型信息系统下进行增量式属性约简的研究,采用邻域知识粒度构造出一种邻域型信息系统的增量式属性约简算法.首先将信息系统的知识粒度在邻域型信息系统下进行推广,提出了邻域知识粒度;然后针对属性增加的情形,研究了邻域知识粒度的增量式更新机制;最后基于这种机制设计出了相应的增量式属性约简算法.实验分析表明所提出的增量式算法具有较高的动态属性约简性能.In order to study incremental attribute reduction in neighborhood information system,an incremental attribute reduction algorithm of neighborhood information system is constructed by using neighborhood knowledge granularity.Firstly,the knowledge granularity of information system is extended in the neighborhood information system,and the neighborhood knowledge granularity is proposed.Secondly,the incremental updating mechanism of neighborhood knowledge granularity is studied for the case of attribute increase.Finally,the corresponding incremental attribute reduction algorithm is designed based on this mechanism.Experimental results show that the incremental algorithm has high dynamic attribute reduction performance.

关 键 词：粗糙集 增量式学习 邻域关系 知识粒度 属性约简 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]