摘要
现有变精度粗糙集属性约简存在缺乏单调性,导致获得的约简结果未必符合预期的问题。为此,首先给出变精度粗糙集极大正域的概念,基于此提出一种新的约简定义,即基于极大正域的属性约简,可以保证获得相对正域最大的约简,同时论证了新约简与其他已有约简间的关系;然后给出基于极大正域变精度粗糙集属性约简一般算法,以及结合遗传算法改进的约简算法;最后,用UCI数据库中的一个数据集验证了该算法的有效性。
There is a lack of monotonicity in existing attribute reduction of variable precision rough sets.This may cause the attribute reduction results miss the expectations.Aimed at this problem,firstly,the definition of maximum positive region in variable precision rough set is proposed.Based on it,the corresponding attribute reduction definition is also given after,which can ensure the maximal reduction of relative positive region.Meanwhile,the relationship between attribute reduction based on maximal positive region and other existing reductions is discussed.Then,a general algorithm for attribute reduction of variable precision rough sets based on maximal positive region is presented.Basing on it,an improved algorithm combined with genetic algorithm is proposed.Finally,a data set in UCI database is used to verify the effectiveness of this algorithm.
作者
张玉
程林海
何莹莹
吕跃进
ZHANG Yu;CHENG Lin-hai;HE Ying-ying;Lü Yue-jin(College of Electrical Engineering,Guangxi University,Nanning 530004,China;College of Mathematics and Information Science,Guangxi University,Nanning 530004,China;Liuzhou Institute of Technology,Liuzhou 545616,China)
出处
《模糊系统与数学》
北大核心
2020年第5期139-149,共11页
Fuzzy Systems and Mathematics
基金
国家自然科学基金资助项目(71361002)。
关键词
变精度粗糙集模型
属性约简
极大正域
遗传算法
决策表
VPRS Model
Attribute Reduction
Maximal Positive Region
Genetic Algorithm
Decision Tables
