Scholarlink

<p>With the introduction of various deep learning models, recent deep multi-view clustering methods have achieved impressive results compared with traditional shallow models. However, due to the lack of effective information interaction, the main modules in these methods are conducted in a separate manner, which inevitably leads to suboptimal solutions. In this paper, a novel <b>B</b>ilevel <b>F</b>eature Learning for <b>M</b>ulti-<b>V</b>iew <b>C</b>lustering with Structural Consistency and Discriminative Enhancement (BF-MVC) is proposed. For the above issue, two different self-guided mechanisms are united in our approach to jointly drive the interaction of view-specific feature learning with the multi-view clustering. First, the structural information focusing on intra-class compactness is obtained from the global feature, which guides the feature learning of each view by manifold regularization. Secondly, the discriminative information focusing on inter-class separation is also extracted from the global feature, which guides the feature learning of each view by minimizing the KL divergence between them. These two kinds of information from global feature can not only represent the semantics of clusters, but also imply consistency among multiple views. Extensive experimental results on several datasets demonstrate the effectiveness of our approach.</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

Bilevel feature learning for multi-view clustering with structural consistency and discriminative enhancement

  • 期刊论文

  • 发表日期：

    2026年3月21日

查看全文
  • Lei Wu,
  • Licheng Ruan,
  • Dongliang Zhang,
  • Xudong Chen,
  • Xinmin Cheng,
  • Hongwei Yin

摘要

With the introduction of various deep learning models, recent deep multi-view clustering methods have achieved impressive results compared with traditional shallow models. However, due to the lack of effective information interaction, the main modules in these methods are conducted in a separate manner, which inevitably leads to suboptimal solutions. In this paper, a novel Bilevel Feature Learning for Multi-View Clustering with Structural Consistency and Discriminative Enhancement (BF-MVC) is proposed. For the above issue, two different self-guided mechanisms are united in our approach to jointly drive the interaction of view-specific feature learning with the multi-view clustering. First, the structural information focusing on intra-class compactness is obtained from the global feature, which guides the feature learning of each view by manifold regularization. Secondly, the discriminative information focusing on inter-class separation is also extracted from the global feature, which guides the feature learning of each view by minimizing the KL divergence between them. These two kinds of information from global feature can not only represent the semantics of clusters, but also imply consistency among multiple views. Extensive experimental results on several datasets demonstrate the effectiveness of our approach.