View-Consistency Learning for Incomplete Multiview Clustering.

Area of Science:

• Machine Learning
• Data Science
• Artificial Intelligence

Background:

• Incomplete multi-view clustering presents challenges due to missing data across different feature sets.
• Existing methods often struggle to effectively integrate information from multiple, incomplete data sources.

Purpose of the Study:

• To develop a general framework for incomplete multi-view clustering.
• To integrate graph learning and spectral clustering for robust representation.
• To address the challenge of missing information in multi-view data.

Main Methods:

• A novel framework integrating graph learning and spectral clustering.
• Tensor low-rank constraint for stable low-dimensional representation.
• Augmented Lagrangian multipliers for algorithm development.
• Tensor Schatten p-norm for improved rank approximation.
• Joint exploitation of consistency and specificity for subspace learning.

Main Results:

• The proposed model effectively encodes complementary information across views.
• It successfully accounts for cluster structures between different incomplete views.
• Extensive experiments show superior performance compared to baseline methods on benchmark datasets.

Conclusions:

• The developed framework offers a robust solution for incomplete multi-view clustering.
• Integrating graph learning, spectral clustering, and tensor methods enhances clustering performance.
• The approach demonstrates significant improvements in handling missing data in multi-view scenarios.