计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (12): 161-169.DOI: 10.3778/j.issn.1002-8331.2003-0288

• 模式识别与人工智能 • 上一篇    下一篇

维度变化的不完备混合型数据增量式属性约简

刘桂枝   

  1. 山西大同大学 物理与电子科学学院,山西 大同 037009
  • 出版日期:2021-06-15 发布日期:2021-06-10

Incremental Attribute Reduction of Incomplete Hybrid Data Based on Dimension Change

LIU Guizhi   

  1. School of Physics and Electronic Science, Shanxi Datong University, Datong, Shanxi 037009, China
  • Online:2021-06-15 Published:2021-06-10

摘要:

增量式属性约简是目前粗糙集理论的重点研究内容。针对不完备混合型信息系统属性变化的情形,提出一种基于正区域方法的增量式属性约简算法。提出了不完备混合型信息系统下正区域的一种等价且高效的计算表达形式,利用这种计算形式分别构造了属性增加和属性减少时正区域地增量式更新,理论证明了其高效性,基于这种增量式更新设计出了相应的增量式属性约简算法。UCI数据集的实验分析表明所提出增量式算法具有一定的有效性和优越性。

关键词: 粗糙集, 属性约简, 不完备混合型信息系统, 正区域, 增量式学习

Abstract:

Incremental attribute reduction is the focus of rough set theory. In this paper, an incremental attribute reduction algorithm based on positive region method is proposed for incomplete hybrid information system. An equivalent and efficient computing expression of positive region in incomplete hybrid information system is proposed. Incremental updating of positive region with attribute increase and attribute decrease is constructed by using this computing form, and its efficiency is proved. Based on this incremental updating, a corresponding incremental attribute reduction algorithm is designed. The experimental analysis of UCI data sets shows that the incremental algorithm is effective and superior.

Key words: rough set, attribute reduction, incomplete hybrid information system, positive region, incremental learning