English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Group fairness and privacy are fundamental aspects in designing trustworthymachine learning models. Previous research has highlighted conflicts betweengroup fairness and different privacy notions. We are the first to demonstratethe alignment of group fairness with the specific privacy notion of attributeprivacy in a blackbox setting. Attribute privacy, quantified by the resistanceto attribute inference attacks (AIAs), requires indistinguishability in thetarget model's output predictions. Group fairness guarantees this therebymitigating AIAs and achieving attribute privacy. To demonstrate this, we firstintroduce AdaptAIA, an enhancement of existing AIAs, tailored for real-worlddatasets with class imbalances in sensitive attributes. Through theoretical andextensive empirical analyses, we demonstrate the efficacy of two standard groupfairness algorithms (i.e., adversarial debiasing and exponentiated gradientdescent) against AdaptAIA. Additionally, since using group fairness results inattribute privacy, it acts as a defense against AIAs, which is currentlylacking. Overall, we show that group fairness aligns with attribute privacy atno additional cost other than the already existing trade-off with modelutility.

On the Alignment of Group Fairness with Attribute Privacy