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

Fairness-aware machine learning has garnered significant attention in recentyears because of extensive use of machine learning in sensitive applicationslike judiciary systems. Various heuristics, and optimization frameworks havebeen proposed to enforce fairness in classification \cite{del2020review} wherethe later approaches either provides empirical results or provides fairnessguarantee for the exact minimizer of the objective function\cite{celis2019classification}. In modern machine learning, Stochastic GradientDescent (SGD) type algorithms are almost always used as training algorithmsimplying that the learned model, and consequently, its fairness properties arerandom. Hence, especially for crucial applications, it is imperative toconstruct Confidence Interval (CI) for the fairness of the learned model. Inthis work we provide CI for test unfairness when a group-fairness-aware,specifically, Disparate Impact (DI), and Disparate Mistreatment (DM) awarelinear binary classifier is trained using online SGD-type algorithms. We showthat asymptotically a Central Limit Theorem holds for the estimated modelparameter of both DI and DM-aware models. We provide online multiplierbootstrap method to estimate the asymptotic covariance to construct online CI.To do so, we extend the known theoretical guarantees shown on the consistencyof the online bootstrap method for unconstrained SGD to constrainedoptimization which could be of independent interest. We illustrate our resultson synthetic and real datasets.

Fairness Uncertainty Quantification: How certain are you that the model is fair?