Improving Robustness of Graph Neural Networks with Heterophily-Inspired Designs

• URL: http://arxiv.org/abs/2106.07767v1
• Date: Mon, 14 Jun 2021 21:39:36 GMT
• Title: Improving Robustness of Graph Neural Networks with Heterophily-Inspired Designs
• Authors: Jiong Zhu, Junchen Jin, Michael T. Schaub, Danai Koutra
• Abstract summary: Many graph neural networks (GNNs) are sensitive to adversarial attacks, and can suffer from performance loss if the graph structure is intentionally perturbed. We show that in the standard scenario in which node features exhibit homophily, impactful structural attacks always lead to increased levels of heterophily. We present two designs -- (i) separate aggregators for ego- and neighbor-embeddings, and (ii) a reduced scope of aggregation -- that can significantly improve the robustness of GNNs.
• Score: 18.524164548051417
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent studies have exposed that many graph neural networks (GNNs) are sensitive to adversarial attacks, and can suffer from performance loss if the graph structure is intentionally perturbed. A different line of research has shown that many GNN architectures implicitly assume that the underlying graph displays homophily, i.e., connected nodes are more likely to have similar features and class labels, and perform poorly if this assumption is not fulfilled. In this work, we formalize the relation between these two seemingly different issues. We theoretically show that in the standard scenario in which node features exhibit homophily, impactful structural attacks always lead to increased levels of heterophily. Then, inspired by GNN architectures that target heterophily, we present two designs -- (i) separate aggregators for ego- and neighbor-embeddings, and (ii) a reduced scope of aggregation -- that can significantly improve the robustness of GNNs. Our extensive empirical evaluations show that GNNs featuring merely these two designs can achieve significantly improved robustness compared to the best-performing unvaccinated model with 24.99% gain in average performance under targeted attacks, while having smaller computational overhead than existing defense mechanisms. Furthermore, these designs can be readily combined with explicit defense mechanisms to yield state-of-the-art robustness with up to 18.33% increase in performance under attacks compared to the best-performing vaccinated model.

Related papers

• HGAttack: Transferable Heterogeneous Graph Adversarial Attack [63.35560741500611]
  Heterogeneous Graph Neural Networks (HGNNs) are increasingly recognized for their performance in areas like the web and e-commerce. This paper introduces HGAttack, the first dedicated gray box evasion attack method for heterogeneous graphs.
  arXiv  Detail & Related papers  (2024-01-18T12:47:13Z)
• Demystifying Structural Disparity in Graph Neural Networks: Can One Size Fit All? [61.35457647107439]
  Most real-world homophilic and heterophilic graphs are comprised of a mixture of nodes in both homophilic and heterophilic structural patterns. We provide evidence that Graph Neural Networks(GNNs) on node classification typically perform admirably on homophilic nodes. We then propose a rigorous, non-i.i.d PAC-Bayesian generalization bound for GNNs, revealing reasons for the performance disparity.
  arXiv  Detail & Related papers  (2023-06-02T07:46:20Z)
• Resisting Graph Adversarial Attack via Cooperative Homophilous Augmentation [60.50994154879244]
  Recent studies show that Graph Neural Networks are vulnerable and easily fooled by small perturbations. In this work, we focus on the emerging but critical attack, namely, Graph Injection Attack. We propose a general defense framework CHAGNN against GIA through cooperative homophilous augmentation of graph data and model.
  arXiv  Detail & Related papers  (2022-11-15T11:44:31Z)
• EvenNet: Ignoring Odd-Hop Neighbors Improves Robustness of Graph Neural Networks [51.42338058718487]
  Graph Neural Networks (GNNs) have received extensive research attention for their promising performance in graph machine learning. Existing approaches, such as GCN and GPRGNN, are not robust in the face of homophily changes on test graphs. We propose EvenNet, a spectral GNN corresponding to an even-polynomial graph filter.
  arXiv  Detail & Related papers  (2022-05-27T10:48:14Z)
• CAP: Co-Adversarial Perturbation on Weights and Features for Improving Generalization of Graph Neural Networks [59.692017490560275]
  Adversarial training has been widely demonstrated to improve model's robustness against adversarial attacks. It remains unclear how the adversarial training could improve the generalization abilities of GNNs in the graph analytics problem. We construct the co-adversarial perturbation (CAP) optimization problem in terms of weights and features, and design the alternating adversarial perturbation algorithm to flatten the weight and feature loss landscapes alternately.
  arXiv  Detail & Related papers  (2021-10-28T02:28:13Z)
• Understanding Structural Vulnerability in Graph Convolutional Networks [27.602802961213236]
  Graph Convolutional Networks (GCNs) are vulnerable to adversarial attacks on the graph structure. We show that structural adversarial examples can be attributed to the non-robust aggregation scheme of GCNs. We show that adopting the aggregation scheme with a high breakdown point could significantly enhance the robustness of GCNs against structural attacks.
  arXiv  Detail & Related papers  (2021-08-13T15:07:44Z)
• Uncertainty-Matching Graph Neural Networks to Defend Against Poisoning Attacks [43.60973654460398]
  Graph Neural Networks (GNNs) are generalizations of neural networks to graph-structured data. GNNs are vulnerable to adversarial attacks, i.e., a small perturbation to the structure can lead to a non-trivial performance degradation. We propose Uncertainty Matching GNN (UM-GNN), that is aimed at improving the robustness of GNN models.
  arXiv  Detail & Related papers  (2020-09-30T05:29:42Z)
• Graph Backdoor [53.70971502299977]
  We present GTA, the first backdoor attack on graph neural networks (GNNs) GTA departs in significant ways: it defines triggers as specific subgraphs, including both topological structures and descriptive features. It can be instantiated for both transductive (e.g., node classification) and inductive (e.g., graph classification) tasks.
  arXiv  Detail & Related papers  (2020-06-21T19:45:30Z)
• Beyond Homophily in Graph Neural Networks: Current Limitations and Effective Designs [28.77753005139331]
  We investigate the representation power of graph neural networks in a semi-supervised node classification task under heterophily or low homophily. Many popular GNNs fail to generalize to this setting, and are even outperformed by models that ignore the graph structure. We identify a set of key designs that boost learning from the graph structure under heterophily.
  arXiv  Detail & Related papers  (2020-06-20T02:05:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.