改进K-means加权自适应多视图数据聚类算法  

Improved K-Means Weighted Adaptive Multi-View Data Clustering Algorithm

作　　者：李丽亚[1] 闫宏印[2] LI li-ya;YAN hong-yin(Taiyuan Institute of Technology,Taiyuan Shanxi 030008,China;Taiyuan University of Technology,Taiyuan Shanxi 030024,China)

机构地区：[1]太原工业学院,山西太原030008 [2]太原理工大学,山西太原030024

出　　处：《计算机仿真》2021年第8期314-317,429,共5页Computer Simulation

基　　金：科技创新项目:结合文本语义特征和深度学习的主观题自动评分研究(2020L0666)。

摘　　要：在如今的大数据时代,视图数据越来越多,由于这些数据表现出明显的多样性和差异性,使得多视图数据聚类成为了大数据的研究重点问题之一。针对多视图数据聚类问题,提出了一种基于改进K-means加权自适应多视图聚类算法。首先,提出加权自适应多视图聚类算法,降低视图同维度变换的复杂性。然后考虑到数据的误差性和离群点问题,对数据条件进行优化处理,把Frobenius范数作为条件进行改进,起到对多视图数据加权的作用。再结合自由度问题,找到多视图数据的最优解,降低目标函数自由度。最后根据K-means优化理论,通过权重系数减少数据对多视图聚类的影响,确定多视图不同簇的聚类中心,从而完成对所有视图数据的优化。基于MATLAB仿真平台,分别对5个数据集采用4种性能评价指标进行仿真验证。实验结果表明,所提出的算法大大减少了运行时间,而且具有较好的聚类性能。In today’s big data era,there are more and more view data,but due to the diversity and differences of these data,how to cluster multi-view data has become one of the key issues today.For this reason,for the problem of multi-view data clustering,an adaptive multi-view clustering algorithm based on improved K-means weight is proposed.First,a weighted adaptive multi-view clustering algorithm was proposed to reduce the complexity of the same-dimensional transformation of views.Then,considering the error and outliers of the data,the data conditions were optimized,and the Frobenius norm was used as the condition to improve the weighting of the multi-view data.Combined with the degree of freedom problem,find the optimal solution of the multi-view data and reduce the degree of freedom of the objective function.Finally,according to the K-means optimization theory,weight coefficients were used to reduce the impact of data on multi-view clustering,determine the clustering centers of different clusters in multi-views,and optimize all view data.Based on the MATLAB simulation platform,four performance evaluation indexes were used to verify the performance of five datasets.Experimental results show that the algorithm proposed in this paper greatly reduces the running time and has better clustering performance.

关 键 词：多视图数据聚类 加权自适应 优化理论 性能指标 数据集 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]