AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

Fairness Constraints: A Flexible Approach for Fair Classification

Muhammad Bilal Zafar, Isabel Valera, Manuel Gomez-Rodriguez, Krishna P. Gummadi

JMLR 2019 · Learning theory and algorithms › Fairness · approach: Causal and policy fairness in decisions · 183 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add covariance disparate impact constraints to logistic regression, with the sensitive feature used at decision timeC-LR vs PP-LR · Adultaccuracy and DI: C-LR accuracy 0.839 and DI of 0.09 vs PP-LR 0.828 accuracy and 0.01 DIMixed
add covariance disparate impact constraints to logistic regression, with matched DIC-LR vs PP-LR · Adultaccuracy and DI: PP-LR achieves an accuracy of 0.840 and DI of 0.07No clear effect
add covariance disparate impact constraints to logistic regressionC-LR vs PP-LR · Bankaccuracy and DI: C-LR accuracy of 0.908 (0.909 for PP-LR) and DI of 0.01 (0.0 for PP-LR)No clear effect
tighten the covariance threshold c toward 0 (ablation)constrained logistic regression · synthetic datafairness vs accuracyMixed
lower the empirical covariance (stricter constraint) (ablation)C-LR and C-SVM · Adult and Bankrelative lossHurt
add disparate impact constraints on RBF-SVM (c=0)RBF SVM · Synthetic RBF-SVMaccuracy: 0.94 -> 0.83 (phi=pi/4); 0.94 -> 0.56 (phi=pi/8)Hurt
constrain both FPR and FNR disparate mistreatment simultaneously (ablation)constrained logistic regression · Synthetic Case IIaccuracyHurt
constrain only FPR (or only FNR) disparate mistreatment (ablation)constrained logistic regression · Synthetic Case IIDM on the other error rateMixed
replace overall accuracy-loss constraint (gamma-LR) with fine-grained constraints keeping non-protected positives (Fine-gamma-LR) (ablation)business necessity formulation · Adult and Bankaccuracy at equal DIHurt
apply multiple/non-binary sensitive attributes constraints (race, gender)C-LR · Adultaccuracy and group acceptanceImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy