Pan, C.-H.; Qu, Y.; Yao, Y.; Wang, M.-J.-S. HybridGNN: A Self-Supervised Graph Neural Network for Efficient Maximum Matching in Bipartite Graphs. Preprints2024, 2024101354. https://doi.org/10.20944/preprints202410.1354.v1
APA Style
Pan, C. H., Qu, Y., Yao, Y., & Wang, M. J. S. (2024). HybridGNN: A Self-Supervised Graph Neural Network for Efficient Maximum Matching in Bipartite Graphs. Preprints. https://doi.org/10.20944/preprints202410.1354.v1
Chicago/Turabian Style
Pan, C., Yao Yao and Mu-Jiang-Shan Wang. 2024 "HybridGNN: A Self-Supervised Graph Neural Network for Efficient Maximum Matching in Bipartite Graphs" Preprints. https://doi.org/10.20944/preprints202410.1354.v1
Abstract
Solving maximum matching problems in bipartite graphs is critical in fields such as computational biology and social network analysis. This study introduces HybridGNN, a novel graph neural network model designed to efficiently address complex matching problems at scale. HybridGNN combines the capabilities of Graph Attention Networks (GAT), GATv2, and Graph SAGE (SAGEConv) layers, integrating techniques like mixed precision training, gradient accumulation, and Jumping Knowledge networks to enhance computational efficiency and performance. Additionally, the incorporation of Graph Isomorphism Networks (GIN) enhances the model's ability to discriminate between structurally different graphs. A time complexity analysis shows that HybridGNN achieves efficient computation across different layers. When evaluated on an email communication dataset, HybridGNN outperformed traditional algorithms such as Hopcroft-Karp, particularly on large and complex graphs. These results demonstrate that HybridGNN offers a powerful and efficient approach for solving maximum matching problems in bipartite graphs, with potential applications in various fields requiring analysis of large-scale and complex graph data.
Keywords
HybridGNN; maximum matching; bipartite Graphs; graph neural networks (GNN); time complexity; graph attention networks (GAT); graph isomorphism networks (GIN); self-supervised learning; mixed precision training
Subject
Computer Science and Mathematics, Artificial Intelligence and Machine Learning
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.