Graph Adversarial Training: Dynamically Regularizing Based on Graph Structure

TL;DR

This paper introduces Graph Adversarial Training (GraphAT), a novel regularization method that enhances the robustness and generalization of graph neural networks by considering the influence of connected examples during adversarial training.

Contribution

The paper proposes GraphAT, a dynamic regularization scheme for graph neural networks that accounts for connected examples, improving robustness and accuracy over standard training methods.

Findings

GraphAT outperforms normal training on GCN by 4.51% in node classification accuracy.

Experiments on citation and knowledge graphs demonstrate the effectiveness of GraphAT.

GraphAT effectively models the impact of connected examples during adversarial training.

Abstract

Recent efforts show that neural networks are vulnerable to small but intentional perturbations on input features in visual classification tasks. Due to the additional consideration of connections between examples (\eg articles with citation link tend to be in the same class), graph neural networks could be more sensitive to the perturbations, since the perturbations from connected examples exacerbate the impact on a target example. Adversarial Training (AT), a dynamic regularization technique, can resist the worst-case perturbations on input features and is a promising choice to improve model robustness and generalization. However, existing AT methods focus on standard classification, being less effective when training models on graph since it does not model the impact from connected examples. In this work, we explore adversarial training on graph, aiming to improve the robustness and generalization of models learned on graph. We propose Graph Adversarial Training (GraphAT), which takes the impact from connected examples into account when learning to construct and resist perturbations. We give a general formulation of GraphAT, which can be seen as a dynamic regularization scheme based on the graph structure. To demonstrate the utility of GraphAT, we employ it on a state-of-the-art graph neural network model --- Graph Convolutional Network (GCN). We conduct experiments on two citation graphs (Citeseer and Cora) and a knowledge graph (NELL), verifying the effectiveness of GraphAT which outperforms normal training on GCN by 4.51% in node classification accuracy. Codes are available via: https://github.com/fulifeng/GraphAT.