Research map / Learning theory and algorithms / Fairness
Decomposing Direct and Indirect Biases in Linear Models Under Demographic Parity Constraint
Bertille Tierny, Arthur Charpentier, Francois Hu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| apply the post-processing fair predictor on top of a linear model | linear regression · CRIME | unfairness (KS): 0.61 -> 0.12 | Improved |
| apply the post-processing fair predictor on top of a linear model | linear regression · LAW | unfairness (KS): .43 -> .07 | Improved |
| apply the post-processing fair predictor on top of a linear model | linear regression · GOSSIS | unfairness (KS): .15 -> .03 | Improved |
| fairness correction costs some accuracy | fair linear model · CRIME | RMSE: 0.15 -> 0.19 | Hurt |
| omit the sensitive attribute (unaware model) | linear regression · CRIME | unfairness (KS): 0.61 -> 0.55 | No clear effect |
| use group-wise standardization with group-specific covariance (ours) instead of intercept-only correction (CS22) | fair linear predictor · CRIME | unfairness (KS): 0.54 -> 0.12 | Improved |
| use ours instead of FS23 (homoscedastic feature assumption) | fair linear predictor · GOSSIS | RMSE: 12.5 ± 3.3 -> 10.4 ± 0.1 | Improved |
| increase direct bias coefficient Ty in synthetic data | ours and CS22 vs FS23 · synthetic data | performance and fairness | Mixed |
| increase Tmean and Tstd of synthetic data | bias decomposition of base model · synthetic data | decomposition components | Improved |
Closest papers by meaning
- Auditing and Enforcing Conditional Fairness via Optimal Transport (AAAI 2025 · similarity 0.89)
- Fairness-Accuracy Trade-Offs: A Causal Perspective (AAAI 2025 · similarity 0.87)
- Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness (NeurIPS 2025 · similarity 0.87)
- FairPFN: A Tabular Foundation Model for Causal Fairness (ICML 2025 · similarity 0.86)
- Causal Bias Detection in Generative Artifical Intelligence (NeurIPS 2026 · similarity 0.86)
- Meta Optimality for Demographic Parity Constrained Regression via Post-Processing (ICML 2025 · similarity 0.86)
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.
