Research map / Federated learning and privacy / Privacy attacks and secure learning
Privacy-Preserving Face Recognition Using Random Frequency Components
Yuxi Mi, Yuge Huang, Jiazhen Ji, Minyi Zhao, Jiaxiang Wu, Xingkun Xu, Shouhong Ding, Shuigeng Zhou
This paper tried a common change: Add data augmentation
add data augmentation (multiple sampled channel subsets per face) · IJB-B (IR-18 + BUPT) · TPR@FPR=1e-4: 81.43 -> 85.77 Improved
What 689 accepted papers report for this change
- 560 improved
- 44 no clear effect
- 26 mixed
- 59 hurt
- Understanding Multimodal Learning: A Loss Landscape Smoothness Perspective ICML 2026 · add data augmentation (jitter, crop, flip) · UPMC-Food101 · accuracy: 92.84 -> 94.93Improved
- AsyncBEV: Cross-modal flow alignment in Asynchronous 3D Object Detection ICLR 2026 · add EMC + asynchronous data augmentation training · nuScenes val, 0 s offset · NDS: 66.7 -> 63.1Hurt
- Towards Understanding Continual Factual Knowledge Acquisition of Language Models: From Theory to Algorithm ICML 2026 · add data augmentation (5 biographies per individual) in pre-training · Biography PT test, unseen templates · EM: 2.71 -> 81.27Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 |
|---|---|---|---|
| train on random channel subsets instead of fixed channel subset (ablation) | PartialFace · IJB-B (IR-18 + BUPT) | TPR@FPR=1e-4: 78.24 -> 85.77 | Improved |
| add data augmentation (multiple sampled channel subsets per face) (ablation) | PartialFace · IJB-B (IR-18 + BUPT) | TPR@FPR=1e-4: 81.43 -> 85.77 | Improved |
| random channels without augmentation versus vanilla fixed channels (ablation) | PartialFace · IJB-B (IR-18 + BUPT) | TPR@FPR=1e-4: 78.24 -> 81.43 | Improved |
| increase augmentation r from 6 to 18 (ablation) | PartialFace · CFP-FP (IR-18 + BUPT) | accuracy: 86.79 -> 91.20 | Improved |
| increase number of permutations n from 1 to 12 (ablation) | PartialFace · CFP-FP (IR-18 + BUPT) | accuracy: 91.55 -> 89.21 | Hurt |
| add permutation of selected channels (P) for privacy | PartialFace · recovery attack | recovery quality | Improved |
| use PartialFace instead of DuetFace under white-box recovery | PartialFace · Recovery attack (white-box) | verification accuracy of recovered images: 96.52 -> 65.35 | Improved |
| apply PartialFace to ArcFace pipeline (random frequency channels) | ArcFace IR-50 · IJB-C | TPR@FPR=1e-4: 95.60 -> 94.93 | Mixed |
| combine PartialFace with CosFace | CosFace IR-18 · CFP-FP | accuracy: 92.89 -> 89.54 | Mixed |
Closest papers by meaning
- Privacy-Preserving Face Recognition Using Trainable Feature Subtraction (CVPR 2024 · similarity 0.91)
- Validating Privacy-Preserving Face Recognition under a Minimum Assumption (CVPR 2024 · similarity 0.90)
- Disguise without Disruption: Utility-Preserving Face De-identification (AAAI 2024 · similarity 0.87)
- Privacy-Preserving Optics for Enhancing Protection in Face De-Identification (CVPR 2024 · similarity 0.86)
- Enhancing Facial Privacy Protection via Weakening Diffusion Purification (CVPR 2025 · similarity 0.86)
- How Private is Private? A Comparative Study for Face De-Identification (NeurIPS 2026 · 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.
