摘要
差别矩阵是很多学者用来计算粗糙集属性约简的一种方法,该方法因其简单、直观、易于理解而得到广泛应用,但是包含在差别矩阵中的冗余元素不仅对属性约简不起作用反而增加存储空间,为消除这些冗余元素提出了一种新的存储结构:二进制链表,通过位运算将差别矩阵中所有的重复元素和父集元素删除,降低差别信息的存储空间。为验证二进制链表的有效性,提出了一种新的属性约简算法。通过UCI数据库中多组数据集对该方法进行测试,并将实验结果与其他算法进行比较,提出的算法可以更快地得到属性约简集并且能够有效地降低存储空间。
The difference matrix is a method used by many scholars to calculate the attribute reduction of rough sets.This method is widely used because it is simple,intuitive and easy to understand,but the redundant elements contained in the difference matrix are not only take no effect on attribute reduction,but also have the problem of high cost of spatial storage.In order to eliminate these redundant elements,a new storage structure is proposed:a binary linked list.All repetitive elements and parent set elements in the discernibility matrix are deleted by bit operation to reduce the storage space of discernibility information.In order to verify the validity of the binary linked list,a new attribute reduction algorithm is proposed.The method is tested by multiple sets of data sets in the UCI database,and the experimental results are compared with other algorithms.The proposed algorithm can get the attribute reduction set faster and can effectively reduce the storage space.
作者
宋剑
蒋瑜
李冬
鲍杨婉莹
SONG Jian;JIANG Yu;LI Dong;BAOYANG Wanying(College of Software Engineering,Chengdu University of Information Technology,Chengdu 610225,China)
出处
《成都信息工程大学学报》
2019年第2期112-117,共6页
Journal of Chengdu University of Information Technology
基金
四川省教育厅重点资助项目(17ZA0071)
关键词
粗糙集
属性约简
差别矩阵
二进制链表
位运算
rough set
attribute reduction
discernibility matrix
binary linked list
bit operations
