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/

Global: Zendy Plus annual

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

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Previous studies have developed fairness methods for biased models thatexhibit discriminatory behaviors towards specific subgroups. While these modelshave shown promise in achieving fair predictions, recent research hasidentified their potential vulnerability to score-based membership inferenceattacks (MIAs). In these attacks, adversaries can infer whether a particulardata sample was used during training by analyzing the model's predictionscores. However, our investigations reveal that these score-based MIAs areineffective when targeting fairness-enhanced models in binary classifications.The attack models trained to launch the MIAs degrade into simplistic thresholdmodels, resulting in lower attack performance. Meanwhile, we observe thatfairness methods often lead to prediction performance degradation for themajority subgroups of the training data. This raises the barrier to successfulattacks and widens the prediction gaps between member and non-member data.Building upon these insights, we propose an efficient MIA method againstfairness-enhanced models based on fairness discrepancy results (FD-MIA). Itleverages the difference in the predictions from both the original andfairness-enhanced models and exploits the observed prediction gaps as attackclues. We also explore potential strategies for mitigating privacy leakages.Extensive experiments validate our findings and demonstrate the efficacy of theproposed method.

When Fairness Meets Privacy: Exploring Privacy Threats in Fair Binary Classifiers through Membership Inference Attacks