Graph Structure Learning for Robust Graph Neural Networks

By Wei Jin, Yao Ma, Xiaorui Liu, Xianfeng Tang, Suhang Wang, Jiliang Tang. Knowledge Discovery and Data Mining, 2020

921 citations recorded by the source corpus.

Graph Neural Networks (GNNs) are powerful tools in representation learning for graphs. However, recent studies show that GNNs are vulnerable to carefully-crafted perturbations, called adversarial attacks. Adversarial attacks can easily fool GNNs in making predictions for downstream tasks. The vulnerability to adversarial attacks has raised increasing concerns for applying GNNs in safety-critical applications. Therefore, developing robust algorithms to defend adversarial attacks is of great significance. A natural idea to defend adversarial attacks is to clean the perturbed graph. It is evident that real-world graphs share some intrinsic properties. For example, many real-world graphs are low-rank and sparse, and the features of two adjacent nodes tend to be similar. In fact, we find that adversarial attacks are likely to violate these graph properties. Therefore, in this paper, we explore these properties to defend adversarial attacks on graphs. In particular, we propose a general framework Pro-GNN, which can jointly learn a structural graph and a robust graph neural network model from the perturbed graph guided by these properties. Extensive experiments on real-world graphs demonstrate that the proposed framework achieves significantly better performance compared with the state-of-the-art defense methods, even when the graph is heavily perturbed. We release the implementation of Pro-GNN to our DeepRobust repository for adversarial attacks and defenses. The specific experimental settings to reproduce our results can be found in https://github.com/ChandlerBang/Pro-GNN.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/2d615707481030d271f2c888b13267b1681f4aa2

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Philip S. Yu
2. Chuan Shi
3. Xiao Wang
4. Alan L. Yuille
5. Jiliang Tang
6. Marinka Zitnik
7. Lichao Sun
8. Xiaohui Shen

Citations in the corpus

• Regional Homogeneity: Towards Learning Transferable Universal Adversarial Perturbations Against Defenses (2019, background): “To mitigate this, large efforts have been devoted to adversarial defense [7, 34, 47, 72], via adversarial training [42, 48, 69, 76, 68, 77, 74], randomized transformation [14, 26, 45, 75] etc.”
• GNNGuard: Defending Graph Neural Networks against Adversarial Attacks (2020, methodology): “In doing so, the GNN is trained to aggregate information from neighbors for every node in each layer, which allows the model to eventually generate representations that capture useful node feature as well as topological structure information [6, 7].”
• Learning to Drop: Robust Graph Neural Network via Topological Denoising (2020, background): “Recent work also shows that graphs with low rank are more robust to network attacks [11, 23].”
• Subgraph Federated Learning with Missing Neighbor Generation (2021, background): “Current robust GNNs focus on the sensitivity towards modifications on node features [16, 17] or adding/removing edges on the graph [18].”
• Mining Latent Structures for Multimedia Recommendation (2021, methodology): “There are three categories of GSL methods: metric learning [5, 22, 37], probabilistic modeling [8, 25, 48], and direct optimization approaches [9, 17, 42].”
• NodeFormer: A Scalable Graph Structure Learning Transformer for Node Classification (2023, background): “approaches [10, 20], Bayesian inference [51] and projected gradient descent [18].”
• Heterogeneous Graph Structure Learning for Graph Neural Networks (2021, methodology): “…learn graph structures for GNNs, graph structure learning (GSL) methods (Franceschi et al. 2019; Jiang et al. 2019; Chen, Wu, and Zaki 2019; Jin et al. 2020) are proposed, most of which parameterize the adjacency matrix and optimize it along with the GNN parameters toward downstream tasks.”
• Adversarial Attack and Defense on Graph Data: A Survey (2018, methodology): “Citation Network Node/Link Citeseer [87] 3,327 4,732 3,703 6 [145], [28], [23], [106], [92], [8], [147], [12], [129], [120], [125], [17], [127], [40], [140], [89], [22], [144], [105] , [57], [148], [9], [52], [53], [107], [36], [134], [96], [20], [54], [29], [56], [31], [76], [77], [146], [59], [49]”
• Node Similarity Preserving Graph Convolutional Networks (2020, methodology): “Similarly, recent research has also demonstrated that graph neural networks are vulnerable to adversarial attacks [14, 30, 44].”