Probing Graph Neural Network Activation Patterns Through Graph Topology
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866910031704752128 |
|---|---|
| author | Tori, Floriano Bini, Lorenzo Sorbi, Marco Marchand-Maillet, Stéphane Ginis, Vincent |
| author_facet | Tori, Floriano Bini, Lorenzo Sorbi, Marco Marchand-Maillet, Stéphane Ginis, Vincent |
| contents | Curvature notions on graphs provide a theoretical description of graph topology, highlighting bottlenecks and denser connected regions. Artifacts of the message passing paradigm in Graph Neural Networks, such as oversmoothing and oversquashing, have been attributed to these regions. However, it remains unclear how the topology of a graph interacts with the learned preferences of GNNs. Through Massive Activations, which correspond to extreme edge activation values in Graph Transformers, we probe this correspondence. Our findings on synthetic graphs and molecular benchmarks reveal that MAs do not preferentially concentrate on curvature extremes, despite their theoretical link to information flow. On the Long Range Graph Benchmark, we identify a systemic \textit{curvature shift}: global attention mechanisms exacerbate topological bottlenecks, drastically increasing the prevalence of negative curvature. Our work reframes curvature as a diagnostic probe for understanding when and why graph learning fails. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_21092 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Probing Graph Neural Network Activation Patterns Through Graph Topology Tori, Floriano Bini, Lorenzo Sorbi, Marco Marchand-Maillet, Stéphane Ginis, Vincent Machine Learning Artificial Intelligence Curvature notions on graphs provide a theoretical description of graph topology, highlighting bottlenecks and denser connected regions. Artifacts of the message passing paradigm in Graph Neural Networks, such as oversmoothing and oversquashing, have been attributed to these regions. However, it remains unclear how the topology of a graph interacts with the learned preferences of GNNs. Through Massive Activations, which correspond to extreme edge activation values in Graph Transformers, we probe this correspondence. Our findings on synthetic graphs and molecular benchmarks reveal that MAs do not preferentially concentrate on curvature extremes, despite their theoretical link to information flow. On the Long Range Graph Benchmark, we identify a systemic \textit{curvature shift}: global attention mechanisms exacerbate topological bottlenecks, drastically increasing the prevalence of negative curvature. Our work reframes curvature as a diagnostic probe for understanding when and why graph learning fails. |
| title | Probing Graph Neural Network Activation Patterns Through Graph Topology |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2602.21092 |