Research map / Label-efficient and robust learning / Recognition and long-tail
Open-Unfairness Adversarial Mitigation for Generalized Deepfake Detection
Zhaoyang Li, Zhu Teng, Baopeng Zhang, Jianping Fan
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 Local Adversarial Learning (ablation) | AdvOU · CDF/DFDC/DFD frame-level avg | AUC: 75.65 -> 80.04 | Improved |
| add Global Relation Debiasing loss (Lro) (ablation) | AdvOU · CDF/DFDC/DFD frame-level avg | AUC: 77.66 -> 80.04 | Improved |
| add Unfairness Relation Amplification (ablation) | AdvOU · CDF/DFDC/DFD frame-level avg | AUC: 77.67 -> 80.04 | Improved |
| use lightweight Unfairness Regulator on frozen detector instead of fully trainable discoverer (ablation) | AdvOU · CDF/DFDC/DFD frame-level avg | AUC: 76.01 -> 80.04 | Improved |
| train unfairness discoverer on both real and fake samples instead of one source (ablation) | AdvOU · CDF/DFDC/DFD frame-level avg | AUC: 78.00 / 77.64 -> 80.04 | Improved |
| plug AdvOU into F3Net | F3Net · CDF, DFDCP, DFDC, DFD avg | AUC: 72.10 -> 77.29 | Improved |
| plug AdvOU into UIA-VIT | UIA-VIT · CDF, DFDCP, DFDC, DFD avg | AUC: 77.14 -> 80.23 | Improved |
| plug AdvOU into EfficientNet-B4 | EfficientNet-B4 · DFD | AUC: 84.95 -> 84.60 | No clear effect |
| AdvOU open-unfairness mitigation on EfficientNet-B3 | EfficientNet-B3 · CDF open-unfairness | FFPR: 88.45 -> 13.6 | Improved |
| AdvOU vs UDD with single ViT | AdvOU ViT · DFDC video-level | AUC: 81.2 (UDD) vs 79.1 (Ours); CDF 93.1 vs 93.9 | Mixed |
Closest papers by meaning
- Preserving Fairness Generalization in Deepfake Detection (CVPR 2024 · similarity 0.91)
- Fair Deepfake Detectors Can Generalize (NeurIPS 2025 · similarity 0.90)
- DAFA: Distance-Aware Fair Adversarial Training (ICLR 2024 · similarity 0.87)
- On the Alignment between Fairness and Accuracy: from the Perspective of Adversarial Robustness (ICML 2025 · similarity 0.87)
- Adversarial Fairness Network (AAAI 2024 · similarity 0.87)
- On the Fairness ROAD: Robust Optimization for Adversarial Debiasing (ICLR 2024 · similarity 0.86)
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.
