Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LSEC: Large-scale spectral ensemble clustering

Hongmin Li, Xiucai Ye, Akira Imakura, Tetsuya Sakurai

TL;DR

This paper tackles the efficiency challenges of large-scale ensemble clustering by introducing LSEC, which combines a divide-and-conquer large-scale spectral clustering approach for generating diverse base clusterings with a bipartite-graph based consensus function. It introduces two acceleration tricks—reusing $K$-nearest neighbors and light-$k$-means—to drastically reduce computation without sacrificing accuracy. The method yields a lower overall complexity than many existing ensemble approaches and demonstrates strong performance on ten large-scale datasets in terms of ACC and NMI, while also achieving faster runtimes. The work offers a practical, scalable framework for ensemble clustering capable of handling datasets with millions of points, with broad implications for applications requiring robust consensus clustering at scale.

Abstract

Ensemble clustering is a fundamental problem in the machine learning field, combining multiple base clusterings into a better clustering result. However, most of the existing methods are unsuitable for large-scale ensemble clustering tasks due to the efficiency bottleneck. In this paper, we propose a large-scale spectral ensemble clustering (LSEC) method to strike a good balance between efficiency and effectiveness. In LSEC, a large-scale spectral clustering based efficient ensemble generation framework is designed to generate various base clusterings within a low computational complexity. Then all based clustering are combined through a bipartite graph partition based consensus function into a better consensus clustering result. The LSEC method achieves a lower computational complexity than most existing ensemble clustering methods. Experiments conducted on ten large-scale datasets show the efficiency and effectiveness of the LSEC method. The MATLAB code of the proposed method and experimental datasets are available at https://github.com/Li- Hongmin/MyPaperWithCode.

LSEC: Large-scale spectral ensemble clustering

TL;DR

This paper tackles the efficiency challenges of large-scale ensemble clustering by introducing LSEC, which combines a divide-and-conquer large-scale spectral clustering approach for generating diverse base clusterings with a bipartite-graph based consensus function. It introduces two acceleration tricks—reusing -nearest neighbors and light--means—to drastically reduce computation without sacrificing accuracy. The method yields a lower overall complexity than many existing ensemble approaches and demonstrates strong performance on ten large-scale datasets in terms of ACC and NMI, while also achieving faster runtimes. The work offers a practical, scalable framework for ensemble clustering capable of handling datasets with millions of points, with broad implications for applications requiring robust consensus clustering at scale.

Abstract

Ensemble clustering is a fundamental problem in the machine learning field, combining multiple base clusterings into a better clustering result. However, most of the existing methods are unsuitable for large-scale ensemble clustering tasks due to the efficiency bottleneck. In this paper, we propose a large-scale spectral ensemble clustering (LSEC) method to strike a good balance between efficiency and effectiveness. In LSEC, a large-scale spectral clustering based efficient ensemble generation framework is designed to generate various base clusterings within a low computational complexity. Then all based clustering are combined through a bipartite graph partition based consensus function into a better consensus clustering result. The LSEC method achieves a lower computational complexity than most existing ensemble clustering methods. Experiments conducted on ten large-scale datasets show the efficiency and effectiveness of the LSEC method. The MATLAB code of the proposed method and experimental datasets are available at https://github.com/Li- Hongmin/MyPaperWithCode.

Paper Structure

This paper contains 22 sections, 21 equations, 2 figures, 9 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Ensemble Clustering
  4. Divide-and-conquer based large-scale spectral clustering algorithm
  5. Proposed Framework
  6. Ensemble Generation based on Large-scale Spectral Clustering
  7. Landmark Selection
  8. Searching $K$-nearest landmarks
  9. Similarity Matrix Construction
  10. Bipartite Graph Partitioning
  11. Consensus Function based on Bipartite Graph Partitioning
  12. Discussion
  13. Computational Complexity Analysis
  14. Relations with Other Methods
  15. Experiments
  16. ...and 7 more sections

Figures (2)

  • Figure 1: An overview of proposed method. Given a dataset, $\frac{m}{q}$ sets of landmarks are first generated, then a set of $K$-nearest neighbors are found for each $R^{(i)}$ and $m$ sparse similarity matrices are constructed, finally the base clusterings are obtained through a bipartite graph partitioning process. The proposed method accelerates the similarity matrix construction by recycling $K$-nearest neighbors and bipartite graph partitioning by applying light-$k$-means.
  • Figure 2: Illustration of the five synthetic datasets. Note that only $0.1\%$ samples of each dataset are plotted.