Robust Optimization as Data Augmentation for Large-scale Graphs
This addresses the challenge of data augmentation for graph data, which is incremental as it builds on existing graph regularizers by focusing on node features rather than topological structures.
The paper tackles the problem of effectively augmenting graph data to improve Graph Neural Networks (GNNs) performance by proposing FLAG, a method that augments node features with adversarial perturbations during training, resulting in enhanced generalization and boosted test-time performance across various tasks.
Data augmentation helps neural networks generalize better by enlarging the training set, but it remains an open question how to effectively augment graph data to enhance the performance of GNNs (Graph Neural Networks). While most existing graph regularizers focus on manipulating graph topological structures by adding/removing edges, we offer a method to augment node features for better performance. We propose FLAG (Free Large-scale Adversarial Augmentation on Graphs), which iteratively augments node features with gradient-based adversarial perturbations during training. By making the model invariant to small fluctuations in input data, our method helps models generalize to out-of-distribution samples and boosts model performance at test time. FLAG is a general-purpose approach for graph data, which universally works in node classification, link prediction, and graph classification tasks. FLAG is also highly flexible and scalable, and is deployable with arbitrary GNN backbones and large-scale datasets. We demonstrate the efficacy and stability of our method through extensive experiments and ablation studies. We also provide intuitive observations for a deeper understanding of our method.