Research map / Learning theory and algorithms / Optimal transport
Optimal Transport under Group Fairness Constraints
Linus Bleistein, Mathieu Dagréou, Francisco Andrade, Thomas Boudou, Aurélien Bellet
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 |
|---|---|---|---|
| relax exact fairness with a penalty (penalized OT) instead of FairSinkhorn | FairSinkhorn · Gaussian synthetic | transport cost vs fairness | Improved |
| enforce exact fairness with FairSinkhorn instead of vanilla Sinkhorn | Sinkhorn · Gaussian synthetic | transport cost | Hurt |
| use penalized OT instead of cost learning | cost learning · Circles synthetic | cost-fairness trade-off | Improved |
| use penalized OT instead of cost learning | cost learning · Gaussian synthetic | cost-fairness trade-off | No clear effect |
| use MLP cost instead of Mahalanobis cost (ablation) | cost learning · Circles synthetic | fairness loss: Mahalanobis cannot go below 10^-2 | Improved |
| increase entropic regularization epsilon in vanilla OT | vanilla entropic OT · Gaussian and Circles synthetic | fairness | No clear effect |
| reuse learned cost instead of solving penalized OT for new samples | penalized OT · Gaussian synthetic (500 test samples) | inference time | Improved |
| use learned cost instead of vanilla OT on unseen samples | vanilla OT · Gaussian synthetic (500 test samples) | fairness loss | Improved |
| use MLP cost instead of Mahalanobis cost on unseen samples (ablation) | cost learning · Gaussian synthetic (500 test samples) | fairness generalization gap | Hurt |
| compare penalized OT with cost learning with 7 income groups | penalized OT · Dating app (semi-synthetic) | cost-fairness trade-off | No clear effect |
Closest papers by meaning
- Promoting Fairness Among Dynamic Agents in Online-Matching Markets under Known Stationary Arrival Distributions (NeurIPS 2024 · similarity 0.85)
- Settling the Maximin Share Fairness for Scheduling among Groups of Machines (ICML 2025 · similarity 0.85)
- Fairness in Social Influence Maximization via Optimal Transport (NeurIPS 2024 · similarity 0.85)
- Strategyproof Mechanisms for Group-Fair Obnoxious Facility Location Problems (AAAI 2024 · similarity 0.84)
- The Price of Opportunity Fairness in Matroid Allocation Problems (NeurIPS 2025 · similarity 0.84)
- Fairness and Efficiency in Online Class Matching (NeurIPS 2024 · similarity 0.84)
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.
