Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Fair Classifiers With Partially Annotated Group Labels
Check 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 confidence-based group labels, randomizing labels for low-confidence samples instead of using only group-labeled data | MFD · UTKFace, 10% group labels | accuracy: 73.88 ($\pm$0.78) -> 82.15 ($\pm$0.58) | Helped |
| Use confidence-based group labels, randomizing labels for low-confidence samples instead of using only group-labeled data | MFD · UTKFace, 10% group labels | Δ_M: 36.00 ($\pm$2.92) -> 27.25 ($\pm$2.77) | Helped |
| Use confidence-based group labels rather than random labels for every unlabeled-group sample | MFD · UTKFace, 10% group labels | Δ_M: 37.25 ($\pm$3.19) -> 27.25 ($\pm$2.77) | Helped |
| Use confidence-based group labels rather than trusting every predicted group label | MFD · UTKFace, 10% group labels | Δ_M: 35.75 ($\pm$4.49) -> 27.25 ($\pm$2.77) | Helped |
| Augment training with target-label-only FairFace data and assign group pseudo-labels with CGL | MFD · UTKFace test set | accuracy: 83.46 -> 84.38 | Helped |
| Augment training with target-label-only FairFace data and assign group pseudo-labels with CGL | MFD · UTKFace test set | Δ_M: 25.00 -> 19.50 | Helped |
| Apply confidence-based pseudo group-label assignment | AD · COMPAS, 10% group labels | Δ_M: 19.57 ($\pm$0.93) -> 13.23 ($\pm$3.82) | Helped |
| Apply confidence-based pseudo group-label assignment rather than trusting all predictions | AD · COMPAS, 10% group labels | accuracy: 65.80 ($\pm$0.38) -> 63.82 ($\pm$1.55) | Hurt |
| Replace predicted labels for low-confidence samples with random labels drawn conditional on the target label (ablation) | CGL · UTKFace, 10% group labels | Δ_M: 35.75 ($\pm$4.49) -> 27.25 ($\pm$2.77) | Helped |
| Add FairFace target-label-only samples to UTKFace training and use CGL to assign their group labels | MFD · UTKFace test set | Δ_A: 16.67 -> 13.00 | Helped |
Closest papers by meaning
- Metric-Agnostic Continual Learning for Sustainable Group Fairness (AAAI 2025 · similarity 0.88)
- Mitigating Label Bias in Machine Learning: Fairness through Confident Learning (AAAI 2024 · similarity 0.87)
- Enhancing Robustness of Last Layer Two-Stage Fair Model Corrections (NeurIPS 2024 · similarity 0.87)
- FRAPPÉ: A Group Fairness Framework for Post-Processing Everything (ICML 2024 · similarity 0.87)
- Enhancing Group Fairness in Online Settings Using Oblique Decision Forests (ICLR 2024 · similarity 0.87)
- FFB: A Fair Fairness Benchmark for In-Processing Group Fairness Methods (ICLR 2024 · similarity 0.87)
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.