Research map / Label-efficient and robust learning / Recognition and long-tail
A Reduction to Binary Approach for Debiasing Multiclass Datasets
Ibrahim Alabdulmohsin, Jessica Schrouff, Oluwasanmi O Koyejo
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 |
|---|---|---|---|
| Use R2B instead of the best listed baseline (DPR) | Adult Income, p=0.0, RF · Adult Income, p=0.0, RF | DP [%]: 1.5±.3 -> 2.2±.2 | Hurt |
| Use R2B instead of the best listed baseline | Adult Income, p=0.0, k-NN · Adult Income, p=0.0, k-NN | DP [%]: 3.0±.7 -> 2.1±.4 | Improved |
| Use R2B instead of the best listed baseline | Adult Income, p=0.5, RF · Adult Income, p=0.5, RF | DP [%]: 1.7±.3 -> 1.0±.2 | Improved |
| Use R2B instead of the best listed baseline | Adult Income, p=0.5, k-NN · Adult Income, p=0.5, k-NN | DP [%]: 2.9±.4 -> 1.1±.2 | Improved |
| Use R2B rather than no debiasing | Adult Income, p=0.5, RF · Adult Income, p=0.5, RF | accuracy [%]: 42.0±0.7 -> 34.0±0.4 | Hurt |
| Use R2B rather than no debiasing | Adult Income, p=0.5, k-NN · Adult Income, p=0.5, k-NN | accuracy [%]: 34.9±0.5 -> 31.1±0.4 | Hurt |
| Use R2B instead of OST | COCO, p=0.0 · COCO, p=0.0 | DP [%]: 2.2±.3 -> 2.4±.4 | Hurt |
| Use R2B instead of OST | COCO, p=0.5 · COCO, p=0.5 | DP [%]: 0.7±.2 -> 1.1±.4 | Hurt |
| Use R2B rather than no debiasing | COCO, p=0.0 · COCO, p=0.0 | accuracy [%]: 40.4±.2 -> 41.3±.2 | Improved |
| Use R2B rather than no debiasing | COCO, p=0.5 · COCO, p=0.5 | accuracy [%]: 43.8±.3 -> 48.1±.4 | Improved |
| Use R2B rather than OST | Dermatology test · Dermatology test | DP [%]: 5.1±1.0 -> 5.0±0.5 | Hurt |
| Use R2B rather than the strongest listed baseline | Dermatology test · Dermatology test | Top-1 accuracy [%]: 58.6±0.5 -> 59.1±0.6 | Improved |
Closest papers by meaning
- Everybody Needs Good Neighbours: An Unsupervised Locality-based Method for Bias Mitigation (ICLR 2023 · similarity 0.86)
- Twofold Debiasing Enhances Fine-Grained Learning with Coarse Labels (AAAI 2025 · similarity 0.84)
- CDMAD: Class-Distribution-Mismatch-Aware Debiasing for Class-Imbalanced Semi-Supervised Learning (CVPR 2024 · similarity 0.84)
- Regression under demographic parity constraints via unlabeled post-processing (NeurIPS 2024 · similarity 0.84)
- Improving Subgroup Robustness via Data Selection (NeurIPS 2024 · similarity 0.84)
- Balanced Data, Imbalanced Spectra: Unveiling Class Disparities with Spectral Imbalance (ICML 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.
