Reconstruction of multiplex networks via graph embeddings

Area of Science:

• Network Science
• Machine Learning
• Data Science

Background:

• Multiplex networks represent systems with multiple interaction types.
• Real-world data often aggregates network layers, obscuring structure.
• Reconstructing hidden multiplexity is crucial for understanding complex systems.

Purpose of the Study:

• To develop a machine-learning framework for reconstructing multiplex network structures.
• To leverage graph embeddings for uncovering hidden network layers.
• To evaluate the framework's performance against existing methods.

Main Methods:

• Utilized graph embeddings to represent networks in geometric space.
• Developed a machine-learning approach for multiplexity reconstruction.
• Conducted systematic experiments on synthetic and real-world networks.

Main Results:

• The proposed framework successfully reconstructs hidden multiplexity.
• Graph embeddings provide effective representations for network layer identification.
• The framework demonstrates superior performance compared to existing reconstruction techniques.

Conclusions:

• Machine learning with graph embeddings offers a powerful solution for multiplex network reconstruction.
• The developed framework enhances the analysis of complex systems with hidden layers.
• This approach advances the field of network science by enabling better data interpretation.