基于Lebesgue和熵度量的不完备邻域决策系统特征选择  

FEATURE SELECTION OF INCOMPLETE NEIGHBORHOOD DECISION SYSTEMBASED ON LEBSGUE AND ENTROPY MEASURE

作　　者：余清[1] 侯丽萍 Yu Qing;Hou Liping(School of Mathematics and Computer Science,Xinyang Vocational and Technical College,Xinyang 464000,Henan,China;School of Information Engineering,Xinyang University of Agriculture and Forestry,Xinyang 464000,Henan,China)

机构地区：[1]信阳职业技术学院数学与计算机科学学院,河南信阳464000 [2]信阳农林学院信息工程学院,河南信阳464000

出　　处：《计算机应用与软件》2023年第8期298-311,共14页Computer Applications and Software

基　　金：河南省科技攻关基金资助项目(172102210120);河南省科技厅科技攻关计划项目(172102210451)。

摘　　要：为了能够处理混合数据集和不完全数据集,并能同时保持原始分类信息,提出一种基于Lebesgue和熵度量的不完备邻域决策系统特征选择方法。提出一种基于邻域容忍度的Lebesgue度量以及基于邻域容忍度的不确定性度量;分析不完全邻域决策系统的不确定性。消除无关特征并设计一种启发式特征选择算法,以提高混合和不完全数据集的分类性能。在多个公共数据集上的实验表明,该方法能够有效地选择最相关的特征,对不完全邻域决策系统具有较好的分类效果。In order to deal with mixed data sets and incomplete data sets,and maintain the original classification information at the same time,a feature selection method for incomplete neighborhood decision system based on Lebesgue and entropy measure is proposed.A Lebesgue measure based on neighborhood tolerance and uncertainty measure based on neighborhood tolerance were proposed.The uncertainty of incomplete neighborhood decision-making system was analyzed.A heuristic feature selection algorithm was designed to improve the classification performance of mixed and incomplete data sets.Experiments on several public datasets show that the proposed feature selection method can effectively select the most relevant features and has strong classification ability for incomplete neighborhood decision-making systems.

关 键 词：特征选择 Lebesgue度量 熵 邻域粗糙集 决策系统 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]