Fast Parameter-Free Multi-View Subspace Clustering With Consensus Anchor Guidance.

Area of Science:

• Machine Learning
• Data Mining
• Computer Vision

Background:

• Multi-view subspace clustering fuses information from multiple sources using graph structures.
• Existing methods face cubic time complexity, limiting scalability.
• Heuristic anchor selection and separate graph construction hinder performance and require hyper-parameter tuning.

Purpose of the Study:

• To develop a novel, efficient, and parameter-free multi-view subspace clustering algorithm.
• To address the scalability and hyper-parameter limitations of current methods.
• To improve clustering performance by jointly optimizing anchor selection and graph construction.

Main Methods:

• Proposes Fast Parameter-free Multi-view Subspace Clustering with Consensus Anchor Guidance (FPMVS-CAG).
• Unifies anchor selection and subspace graph construction into a single optimization framework.
• Achieves linear time complexity and automatic hyper-parameter learning.

Main Results:

• FPMVS-CAG demonstrates linear time complexity, significantly improving efficiency for large datasets.
• The method effectively integrates anchor selection and graph construction, enhancing clustering quality.
• Experimental results validate superior performance and efficiency compared to state-of-the-art methods.

Conclusions:

• FPMVS-CAG offers an effective and efficient solution for large-scale multi-view subspace clustering.
• The parameter-free nature and unified optimization enhance its practical applicability.
• The proposed method outperforms existing approaches in both effectiveness and efficiency.