Home | Publications | BBA+26

Privilege Scores

MCML Authors

Link to Profile Bernd Bischl

Bernd Bischl

Prof. Dr.

Director

Statistical Learning and Data Science

Abstract

Bias-transforming methods of fairness-aware machine learning aim to correct a non-neutral status quo with respect to a protected attribute (PA). We introduce privilege scores (PS) to measure PA-related privilege by comparing the model predictions in the real world with those in a fair world in which the influence of the PA is removed. At the individual level, PS can identify individuals who qualify for affirmative action; at the global level, PS can inform bias-transforming policies. We propose a downstream analysis of PS that allows for statistical significance statements regarding real-world privilege. After presenting estimation methods for PS, we propose privilege score contributions (PSCs), an interpretation method that attributes the origin of privilege to mediating features and direct effects. We provide confidence intervals for both PS and PSCs and propose multi-PS, a multi-objective optimization method for training fairer models by minimizing PS. Experiments on simulated and real-world data demonstrate the broad applicability of our methods and provide novel insights into gender and racial privilege in mortgage and college admissions applications.

inproceedings BBA+26

ACM FAccT 2026

9th ACM Conference on Fairness, Accountability, and Transparency. Montréal, Canada, Jun 25-28, 2026.

Authors

L. Bothmann • P. A. Boustani • J. M. Alvarez • G. Casalicchio • B. Bischl • S. Dandl

Links

DOI

Research Area

 A1 | Statistical Foundations & Explainability

BibTeXKey: BBA+26

#p-bischl
Back to Top