摘要
特征选择是降低数据维度,提高知识获取效率的一类重要数据处理方法.多尺度决策信息系统的最优尺度选择是在一定代价框架下的最优化求解,从而进一步提高数据挖掘的效率.利用图论中的极小顶点覆盖得到相应系统的最优尺度组合,通过构造多尺度系统特征间的邻接关系矩阵,生成该系统诱导的超图.其极小顶点覆盖可视为多尺度信息系统的一个最优尺度组合.在此基础上,提出基于超图的多尺度决策信息系统最优尺度特征求解算法.数值实验表明,该方法能提高数据挖掘时间效率和分类精度,有利于降低数据处理成本.
Feature selection is an important data processing method to reduce the data dimension and improve the efficiency of knowledge acquisition.The optimal scale selection of multi-scale decision information system is the optimal solution under a certain cost framework,to further improve the efficiency of data mining.The optimal scale combination of the corresponding system is obtained by using minimal vertex coverage in graph theory.The hypergraph induced by the system is generated by constructing the adjacency matrix between the features of the multi-scale system,and its minimal vertex coverage can be seen as an optimal scale combination of the multi-scale information system.On this basis,an optimal scaling feature solving algorithm for multi-scale decision information system based on hypergraph is proposed.Numerical experiments show that this method can improve the time efficiency and classification accuracy of data mining,and is beneficial to reduce the cost of data processing.
作者
马周明
黄闽
林国平
施虹艺
MA Zhouming;HUANG Min;LIN Guoping;SHI Hongyi(School of Mathematics and Statistics,Minnan Normal University,Zhangzhou,Fujian 363000,China;Fujian Key Laboratory of Granular Computing and Applications,Minnan Normal University,Zhangzhou,Fujian 363000,China)
出处
《闽南师范大学学报（自然科学版）》
2023年第4期1-15,共15页
Journal of Minnan Normal University：Natural Science
基金
国家自然科学基金项目(11871259,62076088)
福建省自然科学基金项目(2021J01979,2021J01983)
福建省教育厅中青年教师教育科研项目(JAT220211)
闽南师范大学研究生教育改革项目(YJG202209)。
关键词
超最优尺度选择
粒计算
邻接关系
超图
特征选择
optimal scale selection
granular computing
adjacency
hypergraph
feature selection
