基于邻域优势粗糙集的区分度动态属性约简算法
Dynamic attribute reduction algorithm based on neighborhood dominance rough set
张海玉 1贾润亮1
作者信息
- 1. 太原理工大学 财经学院,山西 太原 030024;山西省财政税务专科学校信息学院,山西太原 030024
- 折叠
摘要
为解决动态环境下数值型偏序关系数据的属性约简问题,利用优势粗糙集的区分度提出一种增量式属性约简算法.在数值型信息系统环境下,定义邻域优势区分度度量,通过邻域优势区分度设出一种非增量式属性约简算法;研究和分析对象变化场景下邻域优势区分度进行增量式更新的原理;分别提出数据对象增加和减少情形下数据集属性约简的增量式更新算法.在多个UCI数据集上进行实验验证,实验结果表明,该增量式算法能够有效完成动态数据的属性约简任务.
Abstract
To solve the problem of attribute reduction for numerical ordered data in dynamic environments,an incremental attri-bute reduction algorithm was proposed using the discrimination measurement of dominance rough set.In a numerical information system environment,a neighborhood dominance discrimination measurement was defined,and a non-incremental attribute reduc-tion algorithm was designed based on the neighborhood dominance discrimination measurement.The principle of incremental updating of neighborhood dominance discrimination measurement in object changing scenarios was studied and analyzed.When adding or deleting batch objects in a numerical information system,an incremental updating algorithm for attribute reduction was proposed respectively.Experimental results on several UCI datasets show that the incremental algorithm can effectively accom-plish the task of attribute reduction of dynamic data.
关键词
数值型/偏序关系数据/属性约简/优势粗糙集/邻域关系/区分度/增量式学习Key words
numerical type/ordered data/attribute reduction/dominance rough set/neighborhood relation/discrimination/in-cremental learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据