Scholarlink

<p>Community detection in network analysis has been an attractive research area recently. Here, we consider the problem of estimating community memberships of nodes in an overlapping network. Considering the degree heterogeneity, the degree-corrected mixed membership model is used to generate networks. To estimate network memberships, we propose a spectral clustering approach, termed Spectral Clustering for Mixed Membership, based on an ideal cone structure of the variant for the eigendecomposition of the regularized Laplacian matrix. We show the asymptotic consistency of our algorithm under mild assumptions by providing error bounds for the inferred membership vectors of each node. Numerical results on empirical networks demonstrate that our method enjoys competitive performances with some benchmark methods for estimating network memberships.</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

Regularized spectral clustering in overlapping networks

  • 期刊论文

  • 发表日期：

    2026年7月7日

查看全文
  • Huan Qing,
  • Jingli Wang,
  • Haoran Zhan

摘要

Community detection in network analysis has been an attractive research area recently. Here, we consider the problem of estimating community memberships of nodes in an overlapping network. Considering the degree heterogeneity, the degree-corrected mixed membership model is used to generate networks. To estimate network memberships, we propose a spectral clustering approach, termed Spectral Clustering for Mixed Membership, based on an ideal cone structure of the variant for the eigendecomposition of the regularized Laplacian matrix. We show the asymptotic consistency of our algorithm under mild assumptions by providing error bounds for the inferred membership vectors of each node. Numerical results on empirical networks demonstrate that our method enjoys competitive performances with some benchmark methods for estimating network memberships.