Research map / Label-efficient and robust learning / Recognition and long-tail
Consistent Instance False Positive Improves Fairness in Face Recognition
Xingkun Xu, Yuge Huang, Pengcheng Shen, Shaoxin Li, Jilin Li, Feiyue Huang, Yong Li, Zhen Cui
Check 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 instance FPR penalty to ArcFace-style training | ResNet34 face recognition · RFW trained on BUPT-Balancedface | Avg accuracy (%): 95.30 -> 96.07 | Improved |
| Add instance FPR penalty to ArcFace-style training | ResNet50 face recognition · RFW trained on BUPT-Balancedface | Avg accuracy (%): 95.66 -> 96.52 | Improved |
| Add instance FPR penalty to ArcFace-style training | ResNet100 face recognition · RFW trained on BUPT-Balancedface | Avg accuracy (%): 96.24 -> 96.78 | Improved |
| Add instance FPR penalty to ArcFace-style training | ResNet34 face recognition · RFW trained on BUPT-Globalface | Avg accuracy (%): 95.37 -> 96.56 | Improved |
| Add instance FPR penalty | ResNet34 face recognition · RFW at overall FPR 10^-4 | Bias degree: 208.44 -> 185.91 | Improved |
| Add instance FPR penalty | ResNet34 face recognition · BFW at overall FPR 10^-4 | Bias degree: 2.91 -> 1.67 | Improved |
| Set overall FPR gamma to 10^-4 (ablation) | FPR penalty loss · RFW | Avg accuracy (%): 95.42 -> 96.07 | Improved |
| Set overall FPR gamma to 10^-5 instead of 10^-4 (ablation) | FPR penalty loss · RFW | Std: 0.69 -> 0.91 | Hurt |
| Set exponent p in F(z)=z^p to 2.0 (ablation) | FPR penalty loss with gamma=10^-4 · RFW | Avg accuracy (%): 95.70 -> 96.07 | Improved |
| Set exponent p in F(z)=z^p to 2.0 rather than 1.5 (ablation) | FPR penalty loss with gamma=10^-4 · RFW | Std: 1.05 -> 0.69 | Improved |
Closest papers by meaning
- Instance-Consistent Fair Face Recognition (TPAMI 2025 · similarity 0.92)
- Jointly De-biasing Face Recognition and Demographic Attribute Estimation (ECCV 2020 · similarity 0.89)
- Mitigating Face Recognition Bias via Group Adaptive Classifier (CVPR 2021 · similarity 0.89)
- Benchmarking Algorithmic Bias in Face Recognition: An Experimental Approach Using Synthetic Faces and Human Evaluation (ICCV 2023 · similarity 0.88)
- Mitigating Gender Bias in Face Recognition using the von Mises-Fisher Mixture Model (ICML 2022 · similarity 0.87)
- Assessing Uncertainty in Similarity Scoring: Performance & Fairness in Face Recognition (ICLR 2024 · similarity 0.85)
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.
