Scholarlink

<p>Neighborhood graphs and hierarchical clustering algorithms are effective for characterizing data distributions. However, existing methods face two critical limitations: (1) neighborhood graph construction is dependent on parameters; (2) complex structures are ignored. To address these problems, this paper proposes a hierarchical clustering algorithm based on an adaptive purified neighborhood graph (HC-APNG). The methodology comprises two key innovations: firstly, an adaptive purified neighborhood graph (APNG) is constructed, achieving parameter-free operation while automatically eliminating noise points. Subsequently, HC-APNG also designs a robust subcluster similarity measurement, thereby capturing intricate cluster relationships. The effectiveness of the algorithm in identifying datasets is verified through extensive comparisons on several synthetic and real-world datasets.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hierarchical clustering algorithm based on an adaptive purified neighborhood graph

  • 期刊论文

  • 发表日期：

    2025年9月3日

查看全文
  • Shuang Qin,
  • Ji Feng,
  • Degang Yang,
  • Zhongshang Chen

摘要

Neighborhood graphs and hierarchical clustering algorithms are effective for characterizing data distributions. However, existing methods face two critical limitations: (1) neighborhood graph construction is dependent on parameters; (2) complex structures are ignored. To address these problems, this paper proposes a hierarchical clustering algorithm based on an adaptive purified neighborhood graph (HC-APNG). The methodology comprises two key innovations: firstly, an adaptive purified neighborhood graph (APNG) is constructed, achieving parameter-free operation while automatically eliminating noise points. Subsequently, HC-APNG also designs a robust subcluster similarity measurement, thereby capturing intricate cluster relationships. The effectiveness of the algorithm in identifying datasets is verified through extensive comparisons on several synthetic and real-world datasets.