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.

References in corpus (93)

• Community structure in social and biological networks
• Semi-Supervised Classification with Graph Convolutional Networks
• Deep Learning with Differential Privacy
• Communication-Efficient Learning of Deep Networks from Decentralized Data
• Inductive Representation Learning on Large Graphs
• Graph Convolutional Neural Networks for Web-Scale Recommender Systems
• Interpretable machine learning: definitions, methods, and applications
• Adversarial Attacks on Neural Networks for Graph Data
• GCC: Graph Contrastive Coding for Graph Neural Network Pre-Training
• Explaining Explanations in AI
• How to Explain Individual Classification Decisions
• GNNExplainer: Generating Explanations for Graph Neural Networks
• Pitfalls of Graph Neural Network Evaluation
• From Anecdotal Evidence to Quantitative Evaluation Methods: A Systematic Review on Evaluating Explainable AI
• FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling
• Self-Supervised Graph Transformer on Large-Scale Molecular Data
• Adversarial Attacks on Graph Neural Networks via Meta Learning
• GraphSMOTE: Imbalanced Node Classification on Graphs with Graph Neural Networks
• TUDataset: A collection of benchmark datasets for learning with graphs
• Data Decisions and Theoretical Implications when Adversarially Learning Fair Representations
• XGNN: Towards Model-Level Explanations of Graph Neural Networks
• Higher-Order Explanations of Graph Neural Networks via Relevant Walks
• Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future
• Local Differential Privacy for Deep Learning
• A Survey of Privacy Attacks in Machine Learning
• Iterative Deep Graph Learning for Graph Neural Networks: Better and Robust Node Embeddings
• Fairness Without Demographics in Repeated Loss Minimization
• A Survey of the State of Explainable AI for Natural Language Processing
• Federated Social Recommendation with Graph Neural Network
• Differential Privacy and Machine Learning: a Survey and Review
• Transferring Robustness for Graph Neural Network Against Poisoning Attacks
• CERTIFAI: Counterfactual Explanations for Robustness, Transparency, Interpretability, and Fairness of Artificial Intelligence models
• Learning Adversarially Fair and Transferable Representations
• Fast Gradient Attack on Network Embedding
• EDITS: Modeling and Mitigating Data Bias for Graph Neural Networks
• Graph Unlearning
• GNNGuard: Defending Graph Neural Networks against Adversarial Attacks
• Supervised Community Detection with Line Graph Neural Networks
• How to Find Your Friendly Neighborhood: Graph Attention Design with Self-Supervision
• Trustworthy Graph Neural Networks: Aspects, Methods and Trends
• Certifiable Robustness and Robust Training for Graph Convolutional Networks
• Compositional Fairness Constraints for Graph Embeddings
• Investigating and Mitigating Degree-Related Biases in Graph Convolutional Networks
• Improving Fairness in Graph Neural Networks via Mitigating Sensitive Attribute Leakage
• Quantifying Privacy Leakage in Graph Embedding
• On Explainability of Graph Neural Networks via Subgraph Explorations
• TDGIA:Effective Injection Attacks on Graph Neural Networks
• Evaluating Differentially Private Machine Learning in Practice
• Learning Fair Node Representations with Graph Counterfactual Fairness
• Unnoticeable Backdoor Attacks on Graph Neural Networks
• Single Node Injection Attack against Graph Neural Networks
• GraphLIME: Local Interpretable Model Explanations for Graph Neural Networks
• Discovering Invariant Rationales for Graph Neural Networks
• A Survey of Adversarial Learning on Graphs
• Counterfactual Explanations and Algorithmic Recourses for Machine Learning: A Review
• GIF: A General Graph Unlearning Strategy via Influence Function
• Certifiable Robustness to Graph Perturbations
• Node-Level Membership Inference Attacks Against Graph Neural Networks
• Learning Causally Invariant Representations for Out-of-Distribution Generalization on Graphs
• Fairness without Demographics through Adversarially Reweighted Learning
• PGM-Explainer: Probabilistic Graphical Model Explanations for Graph Neural Networks
• CF-GNNExplainer: Counterfactual Explanations for Graph Neural Networks
• Adversarial Attacks and Defenses on Graphs: A Review, A Tool and Empirical Studies
• Debiasing Graph Neural Networks via Learning Disentangled Causal Substructure
• Unbiased Graph Embedding with Biased Graph Observations
• Interpreting Graph Neural Networks for NLP With Differentiable Edge Masking
• Robust Counterfactual Explanations on Graph Neural Networks
• SpreadGNN: Serverless Multi-task Federated Learning for Graph Neural Networks
• Exploring Edge Disentanglement for Node Classification
• Understanding and Improving Graph Injection Attack by Promoting Unnoticeability
• Towards Fair Graph Neural Networks via Graph Counterfactual
• To be Robust or to be Fair: Towards Fairness in Adversarial Training
• Link Prediction Adversarial Attack
• Generative Causal Explanations for Graph Neural Networks
• Adversarial Attack on Hierarchical Graph Pooling Neural Networks
• GraphDefense: Towards Robust Graph Convolutional Networks
• DeBayes: a Bayesian Method for Debiasing Network Embeddings
• Can Adversarial Network Attack be Defended?
• Exploring the Potential of Large Language Models (LLMs) in Learning on Graphs
• Spectral Augmentation for Self-Supervised Learning on Graphs
• On the Efficiency of the Information Networks in Social Media
• Towards More Practical Adversarial Attacks on Graph Neural Networks
• Model Extraction Attacks on Graph Neural Networks: Taxonomy and Realization
• Fairness-Aware Node Representation Learning
• Learn to Explain Efficiently via Neural Logic Inductive Learning
• Learning Robust Representation through Graph Adversarial Contrastive Learning
• MONET: Debiasing Graph Embeddings via the Metadata-Orthogonal Training Unit
• GARNET: Reduced-Rank Topology Learning for Robust and Scalable Graph Neural Networks
• Randomized Message-Interception Smoothing: Gray-box Certificates for Graph Neural Networks
• Fairness-aware Message Passing for Graph Neural Networks
• A Survey of Graph Unlearning
• A Survey on Privacy in Graph Neural Networks: Attacks, Preservation, and Applications
• Adversarial Attacks on Fairness of Graph Neural Networks

Cited by in corpus (6)

• Towards Multimodal Graph Large Language Model
• Toward Fair Graph Neural Networks Via Dual-Teacher Knowledge Distillation
• Finding Counterfactual Evidences for Node Classification
• Is Your Explanation Reliable: Confidence-Aware Explanation on Graph Neural Networks
• PreGIP: Watermarking the Pretraining of Graph Neural Networks for Deep Intellectual Property Protection
• Unlearning Inversion Attacks for Graph Neural Networks