Certifiably Robust Graph Contrastive Learning
This addresses the security and reliability of unsupervised graph representation learning for applications in domains like social networks or bioinformatics, representing a foundational advancement in robust graph learning.
The paper tackles the vulnerability of Graph Contrastive Learning (GCL) to adversarial attacks by developing the first certifiably robust framework, which includes a unified certification criteria and a novel Randomized Edgedrop Smoothing (RES) technique to ensure provable robustness in GCL models and downstream tasks.
Graph Contrastive Learning (GCL) has emerged as a popular unsupervised graph representation learning method. However, it has been shown that GCL is vulnerable to adversarial attacks on both the graph structure and node attributes. Although empirical approaches have been proposed to enhance the robustness of GCL, the certifiable robustness of GCL is still remain unexplored. In this paper, we develop the first certifiably robust framework in GCL. Specifically, we first propose a unified criteria to evaluate and certify the robustness of GCL. We then introduce a novel technique, RES (Randomized Edgedrop Smoothing), to ensure certifiable robustness for any GCL model, and this certified robustness can be provably preserved in downstream tasks. Furthermore, an effective training method is proposed for robust GCL. Extensive experiments on real-world datasets demonstrate the effectiveness of our proposed method in providing effective certifiable robustness and enhancing the robustness of any GCL model. The source code of RES is available at https://github.com/ventr1c/RES-GCL.