Adversarial Graph Neural Network Benchmarks: Towards Practical and Fair Evaluation

ICLR 2026 Conference Submission21012 Authors

Keywords: graph neural networks, node classification, graph representation learning, evaluation, adversarial machine learning, semi-supervised learning

TL;DR: In this paper, we propose a practical and rigorous evaluation for adversarial GNNs.

Abstract:

Adversarial learning and the robustness of Graph Neural Networks (GNNs) are topics of widespread interest in the machine learning community, as documented by the number of adversarial attacks and defenses designed for these purposes.
While a rigorous evaluation of these adversarial methods is necessary to understand the robustness of GNNs in real-world applications, we posit that many works in the literature do not share the same experimental settings, leading to ambiguous and potentially contradictory scientific conclusions.

In this benchmark, we advocate for standardized, rigorous evaluation practices in adversarial GNN research. We perform a comprehensive re-evaluation of seven widely used attacks and eight recent defenses under both poisoning and evasion scenarios, across six popular graph datasets. Our study spans over 437,000 experiments conducted within a unified framework.

We observe substantial differences in adversarial attack performance when evaluated under a fair and robust procedure. Our findings reveal that previously overlooked factors, such as target node selection and the training process of the attacked model, have a profound impact on attack effectiveness, to the extent of completely distorting performance insights. These results underscore the urgent need for a standardized evaluation framework in adversarial graph machine learning.

Primary Area: datasets and benchmarks

Submission Number: 21012