4.7 Article

Clustering based on local density peaks and graph cut

期刊

INFORMATION SCIENCES
卷 600, 期 -, 页码 263-286

出版社

ELSEVIER SCIENCE INC
DOI: 10.1016/j.ins.2022.03.091

关键词

Clustering; Density peaks; Spectral clustering; Local density; Similarity between trees

资金

  1. National Natural Science Foundation of China [61806170]
  2. Humanities and Social Sciences Fund of Ministry of Education [18XJC72040001]
  3. National Key Research and Development Program of China [2019YFB1706104]

向作者/读者索取更多资源

The article presents a clustering algorithm based on fast search and find of density peaks (DPC), which can achieve highly accurate clustering results when restricted to local neighborhoods. The proposed algorithm captures latent structures in data using family trees and incorporates both distance information and tree structure in the similarity measure. The algorithm outperforms several prominent clustering algorithms in real-world and synthetic datasets.
Clustering by fast search and find of density peaks (DPC) is a widely used and studied clustering algorithm. In this article, we notice that DPC can achieve highly accurate clustering results when restricted to local neighborhoods. Therefore, by investigating density information in local neighborhoods, we propose to capture latent structures in data with family trees, which can reflect density dominations among nearest neighbors of data. A data set will then be partitioned into multiple family trees. In order to obtain the final clustering result, instead of exploiting the error-prone allocation strategy of DPC, we first elaborately design a novel similarity measure for family trees, characterizing not only the distance between data points, but also the structure of trees. Then, we adapt graph cut for the corresponding connection graph to also take global structural information into account. Extensive experiments on both real-world and synthetic data sets show that the proposed algorithm can outperform several prominent clustering algorithms for most of the cases, including the DPC and spectral clustering algorithms and some of their latest variants. We also analyze the robustness of the proposed algorithm w.r.t. hyper-parameters and its time complexity, as well as the necessity of its components through ablation study. (c) 2022 Elsevier Inc. All rights reserved.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.7
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据