Research map / Learning theory and algorithms / Fairness
Optimal Transport of Classifiers to Fairness
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 |
|---|---|---|---|
| Use OTF instead of Norm for equalized-odds fairness regularization | Logistic regression · Adult | AUC-fairness trade-off | Helped |
| Use OTF instead of Norm for equalized-odds fairness regularization | Logistic regression · Diabetes | AUC-PEO violation trade-off | Helped |
| Use OTF instead of Norm for demographic-parity fairness regularization | Logistic regression · Adult, Bank, Dutch, and Diabetes | AUC-fairness trade-off | No clear effect |
| Use OTF instead of Norm for equalized-odds fairness regularization | Logistic regression · Dutch | AUC-PEO violation trade-off | No clear effect |
| Use OTF with multiple sensitive attributes or continuous age | Logistic regression · Adult and Bank | fairness regularization flexibility | Helped |
| Increase smoothing strength to ε = 0.1 or ε = 1 (ablation) | OTF fairness regularizer · Adult PDP, sex | fairness | Hurt |
| Increase smoothing strength to ε = 10 (ablation) | OTF fairness regularizer · Adult PDP, sex | fairness regularizer impact | Hurt |
| Use moderate smoothing strengths ε = 0.001 or ε = 0.01 (ablation) | OTF fairness regularizer · Adult PDP, sex | similarity of group score distributions | Helped |
| Post-process a trained classifier by minimizing adjusted OTF | Logistic regression trained on Adult · Adult PDP, sex | PDP violation | Helped |
| Train with adjusted OTF rather than the unadjusted smoothed OTF cost (ablation) | Probabilistic classifier · Linear fairness constraints | fairness cost for an already-fair classifier | Helped |
Closest papers by meaning
- Testing Group Fairness via Optimal Transport Projections (ICML 2021 · similarity 0.90)
- fairret: a Framework for Differentiable Fairness Regularization Terms (ICLR 2024 · similarity 0.87)
- Post-hoc bias scoring is optimal for fair classification (ICLR 2024 · similarity 0.87)
- Individual Fairness In Strategic Classification (NeurIPS 2025 · similarity 0.86)
- The Fragility of Fairness: Causal Sensitivity Analysis for Fair Machine Learning (NeurIPS 2024 · similarity 0.86)
- Utility-Fairness Trade-Offs and How to Find Them (CVPR 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.