GraphNNK -- Graph Classification and Interpretability
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915766758014976 |
|---|---|
| author | Bolevic, Zeljko Brajovic, Milos Stankovic, Isidora Stankovic, Ljubisa |
| author_facet | Bolevic, Zeljko Brajovic, Milos Stankovic, Isidora Stankovic, Ljubisa |
| contents | Graph Neural Networks (GNNs) have become a standard approach for learning from graph-structured data. However, their reliance on parametric classifiers (most often linear softmax layers) limits interpretability and sometimes hinders generalization. Recent work on interpolation-based methods, particularly Non-Negative Kernel regression (NNK), has demonstrated that predictions can be expressed as convex combinations of similar training examples in the embedding space, yielding both theoretical results and interpretable explanations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_00753 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GraphNNK -- Graph Classification and Interpretability Bolevic, Zeljko Brajovic, Milos Stankovic, Isidora Stankovic, Ljubisa Machine Learning Artificial Intelligence 68T07, 05C50 I.2.6; I.5.1 Graph Neural Networks (GNNs) have become a standard approach for learning from graph-structured data. However, their reliance on parametric classifiers (most often linear softmax layers) limits interpretability and sometimes hinders generalization. Recent work on interpolation-based methods, particularly Non-Negative Kernel regression (NNK), has demonstrated that predictions can be expressed as convex combinations of similar training examples in the embedding space, yielding both theoretical results and interpretable explanations. |
| title | GraphNNK -- Graph Classification and Interpretability |
| topic | Machine Learning Artificial Intelligence 68T07, 05C50 I.2.6; I.5.1 |
| url | https://arxiv.org/abs/2602.00753 |