河北科技大学学报2026，Vol.47Issue(1)：49-59,11.DOI:10.7535/hbkd.2026yx01006
基于最小生成树与统计特征的层次聚类算法
Hierarchical clustering algorithm based on minimum spanning tree and statistical features
摘要
Abstract
To address the limitations of the Chameleon algorithm in terms of parameter sensitivity,noise robustness,and computational efficiency,this study proposed a statistical-MST integrated hierarchical clustering algorithm(SHCA)based on the minimum spanning tree and statistical features.The minimum spanning tree was used to construct a sparse graph,eliminating manual parameter intervention,and the global optimality of the minimum spanning tree was used to avoid false cross cluster connections.The dynamic statistical merging strategy was designed to filter the noise combined with the local distance threshold,and the sub clusters were merged iteratively through the inter cluster connectivity test to ensure the intra cluster compactness and inter cluster separation.Experiment on 20 synthetic datasets and 10 real-world datasets was conducted.The result shows that the proposed SHCA algorithm outperforms existing methods in clustering performance;In cases where performance degradation is observed on certain datasets,the analysis reveals that manifold overlap is the primary contributing factor.Overall,SHCA significantly enhances clustering accuracy and result stability,providing some reference for subsequent research on clustering of large-scale and complex manifold data.关键词
人工智能理论/聚类/层次聚类算法/最小生成树/动态统计合并策略Key words
artificial intelligence theory/clustering/hierarchical clustering algorithm/minimum spanning tree/dynamic sta-tistical merging strategy分类
信息技术与安全科学引用本文复制引用
刘子康,周长杰,姚卫..基于最小生成树与统计特征的层次聚类算法[J].河北科技大学学报,2026,47(1):49-59,11.基金项目
国家自然科学基金(12371462) （12371462）
