arXiv：cs.LG· Yu Song, Zhigang Hua, Yan Xie, Bingheng Li, Jingzhe Liu, Bo Long, Jiliang Tang, Hui Liu·· 3 小时前AI 评分32
SelfAug：面向图数据增强的特征中心方法，被 AAAI 2026 接收
Learning the Latent Structure: A Feature-Centric Approach to Graph Data Augmentation
AI 导读
研究者提出特征中心的图数据增强框架 SelfAug，直接在嵌入向量空间操作、绕过显式结构建模，通过自监督逆掩码过程捕捉观测图与完整图之间的潜在关联，并引入消息正则化器与 bootstrap 策略提升噪声和稀疏监督下的鲁棒性。在覆盖多个领域的十个图数据集上，SelfAug 在归纳与冷启动设置下的准确率和效率均持续优于 SOTA 方法。该工作已被 AAAI 2026 接收。
正文
Abstract:Graph-structured data plays a pivotal role in modeling complex relationships. However, real-world graphs are often incomplete due to data collection and observational constraints, severely limiting the effectiveness of modern graph learning pipelines. While existing Graph Data Augmentation (GDA) methods attempt to refine graph structures for improved downstream performance, they are typically label-dependent, computationally expensive, and inherently transductive, limiting their applicability in practical scenarios. In this work, we present a novel feature-centric graph data augmentation framework that bypasses explicit structure modeling by operating directly in the embedding space. Through a self-supervised inverse masking process, our method captures latent ties between observed and complete graphs, enabling recovery of unobserved structural signals through refined node representations. To enhance robustness under noisy and sparse supervision, we introduce a message regularizer and a bootstrap strategy for effective training and generalization. Evaluated on ten graph datasets spanning multiple domains, our approach, SelfAug, consistently outperforms state-of-the-art methods in both accuracy and efficiency across inductive and cold-start settings, highlighting its potential as a scalable and generalizable solution for real-world graph learning scenarios.
| Comments: | Accepted at AAAI 2026 |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.02517 [cs.LG] |
| (or arXiv:2610.02517v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02517 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yu Song [view email]
[v1]
Thu, 1 Oct 2026 21:38:58 UTC (351 KB)
来源：arXiv：cs.LG · arxiv.org