Weighted tensor-based consistent anchor graph learning for multi-view clustering

Guanghao Du

doi:10.1016/j.neucom.2024.129253

Weighted tensor-based consistent anchor graph learning for multi-view clustering

2025-03-01

0

PRE

AI

DOI：10.1016/j.neucom.2024.129253

Original

Original request for help

Save

Abstract

En 中文

The anchor graph-based multi-view subspace clustering (MVSC) methods have shown promising performance when dealing with large-scale data. However, the consistent information is not fully explored when learning anchor graphs and exploring higher-order relationships, affecting the clustering performance. To this end, we propose a novel multi-view clustering (MVC) model called weighted tensor-based consistent anchor graph learning (WTCAGL). Specifically, we first learn the embedding anchor graphs by projecting the original data into the embedding space. Meanwhile, we learn the consistent anchor graph by concatenating features of different views, which can well capture the consistent relationship between different views to fully explore the cross-view consistent information. Furthermore, we impose the weighted low-rank tensor constraint on the embedding anchor graphs and the consistent anchor graph, which can capture the higher-order relationships between the embedding anchor graph and the consistent anchor graph. Experiments on multi-view datasets validate the effectiveness of the proposed WTCAGL method.

Keywords:

Multi-view clustering

Anchor graph

Consistent information

Weighted low-rank tensor

Weighted tensor-based consistent anchor graph learning for multi-view clustering

Abstract

Journal

Organization