Adversarial Edge

Adversarial edge research focuses on improving the robustness of graph neural networks (GNNs) against attacks that manipulate graph structure by adding, removing, or modifying edges. Current research emphasizes developing methods to detect and mitigate the impact of these adversarial edges, employing techniques like adversarial training, edge dropping (random or guided), and graph structure refinement using both supervised and unsupervised approaches. This work is crucial for enhancing the reliability and security of GNNs in various applications, including those involving sensitive data or critical decision-making, where the integrity of graph data is paramount.