混合型弱标记不完备数据的三支聚类集成仿真  

Three-Branch Cluster Ensemble Simulation of Mixed-Type Weakly Labeled Incomplete Data

作　　者：李慧玲[1] 孙皓 LI Hui-ling;SUN Hao(Computer Department Changzhi University,Changzhi Shanxi 046000,China;Tangshan Iron and Steel Group Co.,Ltd.Education Center,Tangshan Hebei 063000,China)

机构地区：[1]长治学院计算机系,山西长治046000 [2]唐山钢铁集团有限责任公司教育中心,河北唐山063000

出　　处：《计算机仿真》2025年第3期313-316,326,共5页Computer Simulation

基　　金：优创财务管理系统的研发(020/HX1009)。

摘　　要：即使数据存在弱标记和不完备的问题,数据中仍然可能隐藏着有价值的模式和结构。发现数据中的潜在群体或类别,从而挖掘出数据中的有用信息,提出了一种混合型弱标记不完备数据三支聚类集成方法。通过K最近邻填充法对数据的缺失属性展开填充,获得混合型弱标记完备数据集;通过模糊C均值聚类得到数据隶属度,并利用三支聚类集成算法实现混合型弱标记不完备数据的聚类集成。实验结果表明,该方法填充后的数据完整度接近100%,且ARI值在0.85以上,说明所提方法具有较好的数据聚类集成效果,可以有效揭示不同数据类型之间的关系和相互作用。Even if there are weak marks and incomplete data,there may still be valuable patterns and structures hidden in these data.This paper discovers the potential groups or categories in the data,so as to mine useful information in the data,and proposes a three-branch clustering integration method for mixed weakly labeled incomplete data.The missing attributes of the data are filled by the K nearest neighbor filling method to obtain a mixed weakly labeled complete data set.The membership degree of data is obtained by fuzzy C-means clustering,and the clustering integration of mixed weakly labeled incomplete data is realized by using the three-branch clustering integration algorithm.The experimental results show that the data integrity of this method is close to 100%,and the ARI value is above 0.85,which shows that the proposed method has a good data clustering integration effect and can effectively reveal the relationship and interaction between different data types.

关 键 词：混合型弱标记不完备数据 三支聚类集成 最近邻填充法 模糊均值聚类 属性值填充 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]