Learning the Latent Structure: A Feature-Centric Approach to Graph Data Augmentation

2026-10-01

AI Summary

This study addresses structural incompleteness in real-world graph data caused by collection limitations, as well as the reliance of existing augmentation methods on labels, their computational expense, and restriction to transductive settings. To overcome these challenges, we propose SelfAug, a framework that introduces a novel feature-centric augmentation paradigm operating directly in the embedding space to bypass explicit structural modeling. Specifically, it employs a self-supervised inverse masking mechanism to recover unobserved structural signals, complemented by message regularization and bootstrapping strategies to enhance robustness. Extensive experiments across ten multi-domain datasets demonstrate that SelfAug surpasses state-of-the-art methods in both accuracy and efficiency. Furthermore, it achieves effective generalization in inductive learning and cold-start scenarios, validating its potential as a scalable and versatile solution for graph representation learning.

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.

Problem

• Graph Data Augmentation
• Incomplete Graphs
• Label Dependency
• Inductive Learning
• Graph Representation

Innovation

• Graph Data Augmentation
• Feature-Centric
• Self-Supervised Inverse Masking
• Message Regularizer
• Inductive Learning