Research map / Label-efficient and robust learning / Recognition and long-tail
FLAC: Fairness-Aware Representation Learning by Suppressing Attribute-Class Associations
Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, Christos Diou
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 |
|---|---|---|---|
| add the pair selection of Eq. (ablation) | FLAC (Con.) · Biased-MNIST (q=0.999) | accuracy: 28.2 -> 94.1 | Improved |
| add the pair selection of Eq. (ablation) | FairKL-lu · Biased-MNIST (q=0.999) | accuracy: 13.7 -> 73.9 | Improved |
| add the pair selection of Eq. (ablation) | SoftCon · Biased-MNIST (q=0.999) | accuracy: 65.0 -> 84.0 | Improved |
| use both pair terms instead of only same-target different-attribute pairs (ablation) | FLAC (Con.) · Biased-MNIST (q=0.999) | accuracy: 29.1 -> 94.1 | Improved |
| use both pair terms instead of only same-attribute different-target pairs (ablation) | FLAC (CE) · Biased-MNIST (q=0.999) | accuracy: 10.5 -> 89.3 | Improved |
| use student's t kernel instead of cosine (ablation) | FLAC (CE) · Biased-MNIST (q=0.999) | accuracy: 60.4 -> 89.3 | Improved |
| use student's t kernel instead of RBF (ablation) | FLAC (CE) · Biased-MNIST (q=0.999) | accuracy: 83.4 -> 89.3 | Improved |
| use Jeffreys divergence instead of MSE (ablation) | FLAC (CE) · Biased-MNIST (q=0.999) | accuracy: 12.1 -> 89.3 | Improved |
| use Jeffreys divergence instead of KL divergence (ablation) | FLAC (Con.) · Biased-MNIST (q=0.999) | accuracy: 89.2 -> 94.1 | Improved |
| raise alpha from 1 to 1000 on unbiased data (ablation) | FLAC (CE) · Biased-MNIST (q=0.1) | accuracy: 99.4 -> 98.2 | Hurt |
| FLAC-B instead of best prior BL-U | FLAC-B · Corrupted-Cifar10 (q=0.995) | accuracy: 30.0 -> 34.1 | Improved |
| FLAC-B vs best prior BL-U on ImageNet-A | FLAC-B (Con.) · ImageNet-A | accuracy: 35.7 -> 37.9 | Improved |
Closest papers by meaning
- Classes Are Not Equal: An Empirical Study on Image Recognition Fairness (CVPR 2024 · similarity 0.86)
- Leveraging Diffusion Perturbations for Measuring Fairness in Computer Vision (AAAI 2024 · similarity 0.85)
- Kernel-based Equalized Odds: A Quantification of Accuracy-Fairness Trade-off in Fair Representation Learning (NeurIPS 2025 · similarity 0.85)
- Towards Fairness under Label Bias in Image Segmentation: Impact, Measurement and Mitigation (NeurIPS 2026 · similarity 0.85)
- FairDD: Fair Dataset Distillation (NeurIPS 2025 · similarity 0.84)
- Rethinking Fair Representation Learning for Performance-Sensitive Tasks (ICLR 2025 · 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.
