一种基于改进相对邻域区分度的属性约简算法  

An attribute reduction algorithm based on improved relative neighborhood discernibility

作　　者：冯卫兵[1] 孙甜甜 FENG Weibing;SUN Tiantian(College of Science,Xi’an University of Science and Technology,Xi’an 710054,Shaanxi,China)

机构地区：[1]西安科技大学理学院,陕西西安710054

出　　处：《云南大学学报（自然科学版）》2024年第6期1049-1056,共8页Journal of Yunnan University(Natural Sciences Edition)

基　　金：国家自然科学基金(12172280);陕西省自然科学基金(2020CGXNG-013).

摘　　要：弱标记不完备混合型数据是一种常见的数据类型,因此对弱标记不完备混合决策系统进行属性约简,是当前研究的一个热点.首先,通过改进相对邻域区分度的属性重要度定义,构造基于改进相对邻域区分度的增量式更新机制;然后,设计了弱标记不完备混合决策系统中增加属性集的增量式属性约简算法;最后,选取UCI数据库上的8个数据集,将改进的算法与其他同类型的属性约简算法进行对比,实验结果表明,改进的算法具有较高的约简效率和分类性能,从而验证了新算法的可行性.Weakly labeled incomplete mixed data is a common data type,therefore,attribute reduction for weakly labeled incomplete mixed decision systems is a current research hotspot.Firstly,this paper improves the definition of attribute importance based on the relative neighborhood discrimination degree,constructs an incremental updating mechanism based on the improved relative neighborhood discrimination degree.Then,it designs an incremental attribute reduction algorithm for attribute set increase in weakly labeled incomplete mixed decision systems.Finally,we select 8 datasets from the UCI database and compare the improved algorithm with other similar attribute reduction algorithms.The experimental results show that the improved algorithm has higher reduction efficiency and classification performance,thus verifying the feasibility of the proposed algorithm.

关 键 词：属性约简 弱标记数据 相对邻域区分度 增量学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]