New GNN ranks network nodes by betweenness centrality with fewer parameters

Researchers have developed a new Graph Neural Network (GNN) architecture called Degree-Mass Message Passing, designed to efficiently rank nodes by betweenness centrality in large networks. This model leverages the relationship between node importance and multi-hop degree mass, using degree masses as size-invariant features. The approach improves generalization by training on synthetic graphs that mimic real-world network characteristics, including directed and undirected scale-free graphs. Experiments on 14 real-world networks show significant improvements in ranking correlation compared to existing GNN baselines, with a substantial reduction in parameters and faster inference times. AI

IMPACT This new GNN architecture could enable more efficient analysis of large-scale networks, potentially impacting fields like social network analysis, cybersecurity, and scientific literature research.

RANK_REASON Academic paper detailing a new GNN architecture for network analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• Aurora Rossi
• Betweenness Ranking
• Citation Networks
• Degree-Mass Message Passing
• Directed and Undirected Networks
• Graph Neural Network (GNN)
• Hugging Face
• Kendall tau_b
• Social Networks

• paper
• model release
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →