摘要
在实际应用中,决策系统的属性集可能随时间而变化。如何有效地更新约简成为数据挖掘中的重要任务之一。当属性集发生变化时,经典约简算法需要重新计算整个数据。而增量学习充分利用了现有的约简信息,避免了大量的重复计算,从而提高了计算效率。本文针对属性增加和减少的动态数据研究了增量属性约简方法。首先分别设计了属性增加和减少时模糊区分矩阵的更新机制;然后提出了新的属性增加的属性约简算法AIFDM和属性减少的属性约简算法ADFDM.最后,实验结果表明所提的增量算法能够有效的根据属性的增加和减少更新约简,且计算效率提升约1至4.9倍。
In practical application,the attributes of decision system may change over time.How to update reduct effectively has become one of the important tasks in data mining.When the attributes change,the classical reduction algorithm needs to recalculate the data.Incremental learning makes full use of the existing reduct information,so as to avoid a large number of repeated calculations and improve the calculation efficiency.For the dynamic data with attribute increases and decreased,the incremental attribute reduction method AIFDM and ADFDM is studied in this paper.Firstly,the updating mechanism of fuzzy discernibility matrix when attributes are increased and decreased is designed.Then,two new incremental attribute reduction algorithms is proposed.Finally,the experimental results show that the proposed incremental algorithm can effectively update about according to the increase or decrease of attributes,and the the computational efficiency is improved by about 1 to 4.9 times.
作者
刘陈
邓言放
杨田
LIU Chen;DENG Yan-fang;YANG Tian(College of Information Science and Engineering,Hunan Normal University,Changsha 410036,China)
出处
《模糊系统与数学》
北大核心
2023年第1期109-120,共12页
Fuzzy Systems and Mathematics
基金
湖南省自然科学基金优秀青年项目(2021JJ20037)
长沙市杰出创新青年培养计划项目(kq1905031)
关键词
数据挖掘
模糊粗糙集
增量学习
属性约简
模糊区分矩阵
Date Mining
Fuzzy Rough Set
Incremental Learning
Attribute Reduction
Fuzzy Discernibility Matrix
作者简介
刘陈(1999-),女,湖南岳阳人,湖南师范大学研究生,研究方向:粗糙集,数据挖掘等;邓言放(1995-),女,湖南衡阳人,湖南师范大学研究生,研究方向:粗糙集,数据挖掘等;通讯作者:杨田(1984-),女,湖南长沙人,博士,研究方向:粒计算与智能信息处理,粗糙集,模糊集理论,拓扑学等。
