Spectral clustering with distinction and consensus learning on multiple views data.

Area of Science:

• Computer Science
• Data Science
• Machine Learning

Background:

• Multi-view data is common in real-world clustering.
• Existing methods often fail to fully utilize distinct information across views, limiting complementarity.

Purpose of the Study:

• To propose a novel multi-view clustering method that captures both consensus and distinct information.
• To enhance clustering accuracy by exploiting complementary knowledge across different data views.

Main Methods:

• Developed Distinction based Consensus Spectral Clustering (DCSC).
• Incorporated explicit capture of distinct variance from each view.
• Utilized a block coordinate descent algorithm for optimization, with theoretical convergence guarantees.

Main Results:

• DCSC effectively learns a consensus clustering result.
• The method successfully captures and utilizes the distinct variance of each view.
• Experimental results on real-world datasets confirm the method's effectiveness.

Conclusions:

• DCSC offers an improved approach to multi-view clustering by integrating consensus and distinct view information.
• The proposed method enhances clustering performance by better utilizing data complementarity.
• The block coordinate descent algorithm provides an efficient and reliable optimization strategy.