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

Machine learning (ML) models have been deployed for high-stakes applications,e.g., healthcare and criminal justice. Prior work has shown that ML models arevulnerable to attribute inference attacks where an adversary, with somebackground knowledge, trains an ML attack model to infer sensitive attributesby exploiting distinguishable model predictions. However, some prior attributeinference attacks have strong assumptions about adversary's backgroundknowledge (e.g., marginal distribution of sensitive attribute) and pose no moreprivacy risk than statistical inference. Moreover, none of the prior attacksaccount for class imbalance of sensitive attribute in datasets coming fromreal-world applications (e.g., Race and Sex). In this paper, we propose anpractical and effective attribute inference attack that accounts for thisimbalance using an adaptive threshold over the attack model's predictions. Weexhaustively evaluate our proposed attack on multiple datasets and show thatthe adaptive threshold over the model's predictions drastically improves theattack accuracy over prior work. Finally, current literature lacks an effectivedefence against attribute inference attacks. We investigate the impact offairness constraints (i.e., designed to mitigate unfairness in modelpredictions) during model training on our attribute inference attack. We showthat constraint based fairness algorithms which enforces equalized odds acts asan effective defense against attribute inference attacks without impacting themodel utility. Hence, the objective of algorithmic fairness and sensitiveattribute privacy are aligned.

Leveraging Algorithmic Fairness to Mitigate Blackbox Attribute Inference Attacks