Abstract:Graph Contrastive Learning (GCL) relies on semantically consistent graph augmentations, but common local perturbations provide limited control over global structural consistency, motivating a more principled global augmentation strategy. We therefore propose Fractal Graph Contrastive Learning (FractalGCL), a theory-motivated framework that constructs a renormalisation-based augmented graph and introduces a fractal-dimension-aware contrastive loss that penalises unreliable positive views and reweights negative-pair repulsion by finite-scale box-counting discrepancies. However, computing these discrepancies introduces substantial overhead, so we derive and justify a Gaussian surrogate that avoids repeated box-counting on renormalised graphs, yielding about a $61\%$ runtime reduction. Experiments show that FractalGCL serves as an effective frozen-pretraining tool on MalNet-Tiny, achieves strong performance on the standard TUDataset benchmarks, and outperforms the next-best method on real-world urban traffic tasks by $4.51$ percentage points in average accuracy. Code is available at this https URL.
| Comments: | 32 pages, 7 figures |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2505.11356 [cs.LG] |
| (or arXiv:2505.11356v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2505.11356 arXiv-issued DOI via DataCite |
From: Nero Ziyu Li [view email]
[v1]
Fri, 16 May 2025 15:19:10 UTC (1,162 KB)
[v2]
Thu, 22 May 2025 14:40:09 UTC (1,158 KB)
[v3]
Thu, 25 Sep 2025 14:50:34 UTC (1,230 KB)
[v4]
Tue, 12 May 2026 15:36:19 UTC (1,197 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。