Cure-GNN: A Robust Curvature-Enhanced Graph Neural Network Against Adversarial Attacks
- 4 October 2022
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 20 (5) , 4214-4229
Abstract
Graph neural networks (GNNs) are a specialized type of deep learning models on graphs by learning aggregations over neighbor nodes. However, recent studies reveal that the performance of GNNs are severely deteriorated by injecting adversarial examples. Hence, improving the robustness of GNNs is of significant importance. Prior works are devoted to reducing the influence of direct adversaries which are adversarial attacks by positioning a node's one-hop neighbors, yet these approaches are limited in protecting GNNs from indirect adversarial attacks within a node's multi-hop neighbors. In this work, we approach this problem from a new angle by exploring the graph Ricci curvature, which can characterize the relationships of both direct and indirect links from any two nodes' neighborhoods in the Riemannian space. We first investigate the distinguishable properties of adversarial attacks with graph Ricci curvature distribution. Then, a novel defense framework called Cure-GNN is proposed to detect and mitigate adversarial effects. Cure-GNN discerns the distinction between adversarial edges and normal edges via computing curvature, and merges it into the node features reconstructed by a residual learning framework. Extensive experiments over real-world datasets on node classification task demonstrate the efficacy of Cure-GNN and achieves superiority to the state-of-the-arts without incurring high complexity.Keywords
Funding Information
- National Key Research and Development Program of China (2021YFB2700600)
- National Natural Science Foundation of China (62132013, 62102295, 61872082, 61472184)
- Key Research and Development Programs of Shaanxi (2021ZDLGY06-03)
- Guangdong Leading Talent Program (2016LJ06D658)
- Guangdong Innovation Team Program (2018KCXTD030)
This publication has 28 references indexed in Scilit:
- Hiding individuals and communities in a social networkNature Human Behaviour, 2018
- Web service composition using the bidirectional Dijkstra algorithmIEEE Latin America Transactions, 2016
- Deep Residual Learning for Image Recognition2016
- Ricci curvature of graphsTohoku Mathematical Journal, 2011
- Ricci curvature for metric-measure spaces via optimal transportAnnals of Mathematics, 2009
- Ricci curvature of Markov chains on metric spacesJournal of Functional Analysis, 2009
- Optimal Transport2009
- Collective Classification in Network DataAI Magazine, 2008
- ArnetMiner2008
- The political blogosphere and the 2004 U.S. election2005