Enhancing Fairness in Unsupervised Graph Anomaly Detection through Disentanglement
Abstract: Graph anomaly detection (GAD) is becoming increasingly crucial in various applications, ranging from financial fraud detection to fake news detection. However, current GAD methods largely overlook the fairness problem, which might result in discriminatory decisions
skewed toward certain demographic groups defined on sensitive attributes (e.g., gender). This greatly limits the applicability of these methods in real-world scenarios in light of societal and ethical restrictions. To address this critical gap, we make the first attempt
to integrate fairness with utility in GAD decision-making. Specifically, we devise a novel DisEntangle-based FairnEss-aware aNomaly Detection framework on the attributed graph, named DEFEND. DEFEND first introduces disentanglement in GNNs to capture informative yet sensitive-irrelevant node representations, effectively reducing bias inherent in graphrepresentation learning. Besides, to alleviate discriminatory bias in evaluating anomalies, DEFEND adopts a reconstruction-based method, which concentrates solely on node attributes and avoids incorporating biased graph topology. Additionally, given the inherent association between sensitive-relevant and -irrelevant attributes, DEFEND further constrains the correlation between the reconstruction error and predicted sensitive attributes. Empirical evaluations on real-world datasets reveal that DEFEND performs effectively in GAD and significantly enhances fairness compared to state-of-the-art baselines. Our code is available at https://github.com/AhaChang/DEFEND.
Submission Length: Regular submission (no more than 12 pages of main content)
Assigned Action Editor: ~Sheng_Li3
Submission Number: 3739
Loading