Research map / Learning theory and algorithms / Fairness
Generalized Disparate Impact for Configurable Fairness Solutions in ML
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Enforce the GeDI-V1 constraint with Moving Targets | Gradient Boosting on Communities & Crimes with a binary protected attribute · Communities & Crimes, binary protected attribute | R2 (val): 0.63 ± 0.02 -> 0.54 ± 0.02 | Hurt |
| Enforce the GeDI-V1 constraint with semantics-based regularization | Neural Network on Communities & Crimes with a binary protected attribute · Communities & Crimes, binary protected attribute | GeDI-V1 % (val): 0.97 ± 0.12 -> 0.24 ± 0.07 | Helped |
| Enforce the GeDI-V1 constraint with Moving Targets | Gradient Boosting on Adult with a binary protected attribute · Adult, binary protected attribute | Accuracy (val): 0.86 ± 0.00 -> 0.85 ± 0.00 | Hurt |
| Enforce the GeDI-V1 constraint with semantics-based regularization | Neural Network on Adult with a binary protected attribute · Adult, binary protected attribute | GeDI-V1 % (val): 1.03 ± 0.08 -> 0.17 ± 0.03 | Helped |
| Enforce the coarse GeDI-V5 constraint with Moving Targets | Gradient Boosting on Communities & Crimes with a continuous protected attribute · Communities & Crimes, continuous protected attribute | R2 (val): 0.63 ± 0.02 -> 0.49 ± 0.04 | Hurt |
| Enforce the fine-grained GeDI-V5 constraint with Moving Targets | Gradient Boosting on Communities & Crimes with a continuous protected attribute · Communities & Crimes, continuous protected attribute | GeDI-V5 % (val): 2.09 ± 0.50 -> 0.55 ± 0.24 | Helped |
| Enforce the coarse GeDI-V5 constraint with Moving Targets | Gradient Boosting on Adult with a continuous protected attribute · Adult, continuous protected attribute | Accuracy (val): 0.80 ± 0.01 -> 0.83 ± 0.01 | Helped |
| Enforce the fine-grained GeDI-V5 constraint with Moving Targets | Neural Network on Adult with a continuous protected attribute · Adult, continuous protected attribute | GeDI-V5 % (val): 4.12 ± 0.21 -> 0.51 ± 0.29 | Helped |
| Increase polynomial kernel order (ablation) | GeDI-Fine · Preprocessing experiments, binned DIDI-n | binned DIDI | Helped |
| Use fine-grained rather than coarse-grained GeDI constraints (ablation) | GeDI preprocessing · Preprocessing experiments, binned DIDI-n | binned DIDI consistency across bin numbers | Helped |
Closest papers by meaning
- Fairness with Censorship: Bridging the Gap between Fairness Research and Real-World Deployment (AAAI 2024 · similarity 0.89)
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.88)
- Intersectional Fairness via Mixed-Integer Optimization (NeurIPS 2026 · similarity 0.87)
- Fairness Shields: Safeguarding against Biased Decision Makers (AAAI 2025 · similarity 0.87)
- The Fragility of Fairness: Causal Sensitivity Analysis for Fair Machine Learning (NeurIPS 2024 · similarity 0.86)
- Causal Adversarial Perturbations for Individual Fairness and Robustness in Heterogeneous Data Spaces (AAAI 2024 · 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.