Smoothness Regularized Multiview Subspace Clustering With Kernel Learning.

Area of Science:

• Machine Learning
• Data Mining
• Computer Vision

Background:

• Multiview subspace clustering methods often assume linear relationships, limiting their ability to capture complex data patterns.
• Existing approaches may fail to preserve the inherent locality of data in the learned representations, impacting clustering performance.

Purpose of the Study:

• To propose a novel multiview subspace clustering method, Smoothness Regularized Multiview Subspace Clustering with Kernel Learning (SMSCK).
• To address limitations of existing methods by capturing nonlinear relationships and preserving data locality.

Main Methods:

• Employs kernel learning to map data into a high-dimensional space, enabling the modeling of nonlinear relationships.
• Integrates smoothness regularization to preserve the local structure of the original feature space within the learned affinity representation.

Main Results:

• Theoretical analysis confirms the proposed model's grouping effect and guarantees an optimal solution.
• Experimental results on image and document datasets demonstrate superior performance compared to state-of-the-art methods.

Conclusions:

• SMSCK effectively captures nonlinearities and preserves data locality in multiview subspace clustering.
• The method offers a robust and effective approach for complex data clustering tasks.