Graph Vulnerability and Robustness: A Survey

Freitas, Scott; Yang, Diyi; Kumar, Srijan; Tong, Hanghang; Chau, Duen Horng

doi:10.1109/tkde.2022.3163672

Cited by 51 publications

(24 citation statements)

References 136 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where T (u) is the number of triangles through node u and d(u) is the degree of node u [58]. A recent survey [60] on graph vulnerability and robustness suggests several alternative metrics based on graph measures, adjacency measures, and Laplacian measures. We select maximum clustering coefficient as one of our attack strategy, as nodes with highest clustering coefficient tend to have a more connected local neighborhood where the malicious updates can propagate.…”

Section: Stronger Structural Graph Attackmentioning

confidence: 99%

Backdoor Attacks in Peer-to-Peer Federated Learning

Yar¹,

Nita-Rotaru²,

Oprea³

2023

Preprint

View full text Add to dashboard Cite

We study backdoor attacks in peer-to-peer federated learning systems on different graph topologies and datasets. We show that only 5% attacker nodes are sufficient to perform a backdoor attack with 42% attack success without decreasing the accuracy on clean data by more than 2%. We also demonstrate that the attack can be amplified by the attacker crashing a small number of nodes. We evaluate defenses proposed in the context of centralized federated learning and show they are ineffective in peer-to-peer settings. Finally, we propose a defense that mitigates the attacks by applying different clipping norms to the model updates received from peers and local model trained by a node.

show abstract

Section: Stronger Structural Graph Attackmentioning

confidence: 99%

Backdoor Attacks in Peer-to-Peer Federated Learning

Yar¹,

Nita-Rotaru²,

Oprea³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In this section, we summarise recent studies on GNN robustness. Ideally, a robust GNN should be capable of remaining stable under circumstances of both adversarial attacks and random errors (e.g., unexpected omissions in the graph data [85]). While research has shown that random failures are often less severe [86], most current studies focus on robustness when dealing with adversarial attacks [81], [82], which is critical due to its detrimental impacts on GNNbased applications.…”

Section: Robustness Of Gnnsmentioning

confidence: 99%

Trustworthy Graph Neural Networks: Aspects, Methods and Trends

Zhang¹,

Wu²,

Yuan³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Graph neural networks (GNNs) have emerged as a series of competent graph learning methods for diverse real-world scenarios, ranging from daily applications like recommendation systems and question answering to cutting-edge technologies such as drug discovery in life sciences and n-body simulation in astrophysics. However, task performance is not the only requirement for GNNs. Performance-oriented GNNs have exhibited potential adverse effects like vulnerability to adversarial attacks, unexplainable discrimination against disadvantaged groups, or excessive resource consumption in edge computing environments. To avoid these unintentional harms, it is necessary to build competent GNNs characterised by trustworthiness. To this end, we propose a comprehensive roadmap to build trustworthy GNNs from the view of the various computing technologies involved. In this survey, we introduce basic concepts and comprehensively summarise existing efforts for trustworthy GNNs from six aspects, including robustness, explainability, privacy, fairness, accountability, and environmental well-being. Additionally, we highlight the intricate cross-aspect relations between the above six aspects of trustworthy GNNs. Finally, we present a thorough overview of trending directions for facilitating the research and industrialisation of trustworthy GNNs.

show abstract

“…Studies of network resilience and robustness can be found in the engineering and computer science literature. However, much of the work focuses exclusively on measurements of network topology properties (see Freitas et al (2022) for a recent overview) or is based on models of lateral network movements that do not capture the infection and recovery dynamics of risk contagion (see Chen, Tong, and Ying (2018) or Freitas et al (2020)). Moreover, resilience building is studied only from a network perspective and not in a regulatory framework.…”

Section: Literaturementioning

confidence: 99%

Building Resilience in Cybersecurity -- An Artificial Lab Approach

Awiszus¹,

Yannick²,

Lüttringhaus³

et al. 2022

Preprint

View full text Add to dashboard Cite

Based on classical contagion models we introduce an artificial cyber lab: the digital twin of a complex cyber system in which possible cyber resilience measures may be implemented and tested. Using the lab, in numerical case studies, we identify two classes of measures to control systemic cyber risks: security-and topology-based interventions. We discuss the implications of our findings on selected real-world cybersecurity measures currently applied in the insurance and regulation practice or under discussion for future cyber risk control. To this end, we provide a brief overview of the current cybersecurity regulation and emphasize the role of insurance companies as private regulators. Moreover, from an insurance point of view, we provide first attempts to design systemic cyber risk obligations and to measure the systemic risk contribution of individual policyholders.

show abstract

Graph Vulnerability and Robustness: A Survey

Cited by 51 publications

References 136 publications

Backdoor Attacks in Peer-to-Peer Federated Learning

Backdoor Attacks in Peer-to-Peer Federated Learning

Trustworthy Graph Neural Networks: Aspects, Methods and Trends

Building Resilience in Cybersecurity -- An Artificial Lab Approach

Contact Info

Product

Resources

About