A weighted prior tensor train decomposition method for community detection in multi-layer networks.

Area of Science:

• Network Analysis
• Data Science
• Computer Science

Background:

• Community detection is crucial in network analysis.
• Existing methods struggle with high-dimensional multi-layer networks and auxiliary information.
• There's a need for advanced techniques to improve community detection accuracy.

Purpose of the Study:

• To propose a novel approach for multi-layer network community detection.
• To address limitations of existing methods in handling high-dimensional data and auxiliary information.
• To enhance community detection accuracy by leveraging inter-community connections and manifold learning.

Main Methods:

• Weighted Prior Tensor Training Decomposition (WPTTD) for high-dimensional data management.
• Tensor feature optimization techniques within the WPTTD framework.
• Integration of Common Community Manifold Learning (CCML) to preserve community structure and utilize comprehensive network information.
• Construction of prior information using a weighted flattened network to explore inter-community connections.

Main Results:

• WPTTD effectively manages high-dimensional data in multi-layer networks.
• The integration of CCML enhances the preservation of cohesive community structures.
• Experimental results demonstrate superior performance compared to mainstream multi-layer network community detection algorithms.
• The method successfully leverages auxiliary information among communities to boost detection accuracy.

Conclusions:

• WPTTD offers a robust solution for community detection in complex multi-layer networks.
• The proposed method overcomes key limitations of existing techniques.
• WPTTD provides a promising direction for future research in network analysis and community detection.