Topology Attack and Defense for Graph Neural Networks: An Optimization Perspective

• DOI: 10.48550/arxiv.1906.04214
• Publication Date: 2019-01-01
• Authors: Kaidi Xu; Hongge Chen; Sijia Liu; Pin‐Yu Chen; Tsui-Wei Weng; Mingyi Hong; Xue Lin
• Keywords: Robustness; Deep Neural Networks; Robust Optimization

Graph neural networks (GNNs) which apply the deep to graph data have achieved significant performance for task of semi-supervised node classification. However, only few work has addressed adversarial robustness GNNs. In this paper, we first present a novel gradient-based attack method that facilitates difficulty tackling discrete data. When comparing current attacks on GNNs, results show by perturbing small number edge perturbations, including addition and deletion, our optimization-based can lead noticeable decrease in classification performance. Moreover, leveraging attack, propose training Our yields higher against both different gradient based greedy methods without sacrificing accuracy original graph.