pipette
ENEnglish

Empirical Auditing of Edge-Private Graph Generators

Anum Fatima, Stratis Limnios, James Adams, Lukasz Szpruch, Carsten Maple, Gesine Reinert, Andrew Elliott

Preprint

En palabras de los autores

We empirically audit privacy leakage by testing whether outputs from edge-neighbouring inputs remain distinguishable, using statistically valid lower bounds on the privacy loss witnessed by our attacks. Our framework compares direct-edge, local-structural, and GNN-based attacks through the geometry surrounding a target edge. Experiments across two generators and two networks show that privacy leakage is both mechanism- and network-dependent, with learned representations revealing information not captured by conventional local statistics.

Resultado principalLimitación que admiten los autores

Apareció: miércoles, 23 de septiembre. arXiv. Preprint, todavía sin revisión por pares.