Multi-view Subspace Adaptive Learning via Autoencoder and Attention

TL;DR

This paper introduces MSALAA, a novel multi-view subspace learning method combining autoencoders and attention mechanisms to improve clustering performance by effectively fusing multiple data views.

Contribution

The paper proposes a new multi-view subspace learning approach that aligns self-representations across views using autoencoders and attention, enhancing non-linear fitting and view fusion.

Findings

Significant improvement over baseline methods on six datasets

Effective alignment of multi-view self-representations

Enhanced non-linear data modeling capabilities

Abstract

Multi-view learning can cover all features of data samples more comprehensively, so multi-view learning has attracted widespread attention. Traditional subspace clustering methods, such as sparse subspace clustering (SSC) and low-ranking subspace clustering (LRSC), cluster the affinity matrix for a single view, thus ignoring the problem of fusion between views. In our article, we propose a new Multiview Subspace Adaptive Learning based on Attention and Autoencoder (MSALAA). This method combines a deep autoencoder and a method for aligning the self-representations of various views in Multi-view Low-Rank Sparse Subspace Clustering (MLRSSC), which can not only increase the capability to non-linearity fitting, but also can meets the principles of consistency and complementarity of multi-view learning. We empirically observe significant improvement over existing baseline methods on six real-life datasets.