A Comprehensive Survey on Trustworthy Graph Neural Networks: Privacy, Robustness, Fairness, and Explainability

Abstract

Graph Neural Networks (GNNs) have made rapid developments in the recent years. Due to their great ability in modeling graph-structured data, GNNs are vastly used in various applications, including high-stakes scenarios such as financial analysis, traffic predictions, and drug discovery. Despite their great potential in benefiting humans in the real world, recent study shows that GNNs can leak private information, are vulnerable to adversarial attacks, can inherit and magnify societal bias from training data and lack interpretability, which have risk of causing unintentional harm to the users and society. For example, existing works demonstrate that attackers can fool the GNNs to give the outcome they desire with unnoticeable perturbation on training graph. GNNs trained on social networks may embed the discrimination in their decision process, strengthening the undesirable societal bias. Consequently, trustworthy GNNs in various aspects are emerging to prevent the harm from GNN models and increase the users' trust in GNNs. In this paper, we give a comprehensive survey of GNNs in the computational aspects of privacy, robustness, fairness, and explainability. For each aspect, we give the taxonomy of the related methods and formulate the general frameworks for the multiple categories of trustworthy GNNs. We also discuss the future research directions of each aspect and connections between these aspects to help achieve trustworthiness.

Privacy explanations – A means to end-user trust

“…The works [71] and [72] deal with privacy, fairness and explainability in the context of graph neural networks (GNNs). The authors mention in [71] a positive interaction between explainability and privacy and that there is also a connection with trust.…”

“…The works [71] and [72] deal with privacy, fairness and explainability in the context of graph neural networks (GNNs). The authors mention in [71] a positive interaction between explainability and privacy and that there is also a connection with trust. However, [72] refers to possible problems in terms of privacy, when providing explanations for whitebox GNNs, since the model parameters are accessible.…”

HetGLM: Lateral Movement Detection by Discovering Anomalous Links with Heterogeneous Graph Neural Network

“…For future work, we plan to further improve the detection ability of HetGLM by taking temporal information into account. Another future direction is to analyze the certifiable robustness [25] of MADR to provide more reliable detection results.…”

HP-GMN: Graph Memory Networks for Heterophilous Graphs

“…Graph Neural Networks have shown great ability in various graph tasks such as node classification [1], [2], link prediction [3], [4], and graph classification [5]. Generally, the success of GNNs relies on the message-passing mechanism, where a node representation will be updated by aggregating the representations of its neighbors.…”