Researchers audit differentially private graph releases
Original title: Practical Auditing of Differentially Private Graph Release Mechanisms: Methodology, Toolkit, and Case Study
New toolkit falsifies claimed privacy guarantees in Node-DP mechanism.
A methodology and open-source Python toolkit were developed for auditing differential privacy mechanisms on graph data. Two graph-specific audit design problems were identified. The toolkit was applied to a state-of-the-art Node-DP implementation. Certified lower bounds on the privacy parameter ε exceeded the claimed value across synthetic and real-world canaries.
Unconfirmed: No specific implementation details beyond the audited mechanism.
- Canary construction and witness statistic selection are key.
- Violations persist under stronger input domain restrictions.
- Diverse statistics affect violation detection.
From the original article
1 min read at arxiv.orgDifferential Privacy (DP) has become a prominent approach for protecting graph data releases, yet empirical auditing methodologies have largely been developed for tabular datasets and machine learning pipelines rather than graph-structured outputs.
By Nicholas D'Silva · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.