Research map / Label-efficient and robust learning / Knowledge distillation
Evaluation-Oriented Knowledge Distillation for Deep Face Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| distill with evaluation-oriented critical-pair constraints | MobileFaceNet · CFP-FP | verification accuracy: 91.66 -> 94.33 | Helped |
| distill with evaluation-oriented critical-pair constraints | IR18 · CFP-FP | verification accuracy: 94.60 -> 95.31 | Helped |
| use EKD rather than MobileFaceNet baseline | MobileFaceNet · IJB-C, FPR=1e-4 | TPR: 89.13 -> 90.48 | Helped |
| use EKD rather than MobileFaceNet baseline | MobileFaceNet · MegaFace refined rank-1 | identification accuracy: 90.91 -> 91.02 | Helped |
| use EKD rather than MobileFaceNet baseline | MobileFaceNet · MegaFace refined rank-1 | identification accuracy: 91.70 -> 91.02 | Hurt |
| set sigmoid temperature to 0.1 rather than 0.01 (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 94.33 -> 93.33 | Hurt |
| set sigmoid temperature to 0.001 rather than 0.01 (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 94.33 -> 93.29 | Hurt |
| select 1000 rather than 2000 hard negative pairs (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 94.33 -> 93.66 | Hurt |
| select 5000 rather than 2000 hard negative pairs (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 94.33 -> 93.74 | Hurt |
| replace hard negative mining with random selection at N=2000 (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 94.33 -> 94.19 | No clear effect |
| use 3 rather than 6 thresholds (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 94.33 -> 93.57 | Hurt |
| use the relaxed rank-based loss rather than absolute-distance loss (ablation) | EKD MobileFaceNet · CFP-FP | verification accuracy: 91.99 -> 94.33 | Helped |
Closest papers by meaning
- VkD: Improving Knowledge Distillation using Orthogonal Projections (CVPR 2024 · similarity 0.87)
- Over-parameterized Student Model via Tensor Decomposition Boosted Knowledge Distillation (NeurIPS 2024 · similarity 0.86)
- ScaleKD: Strong Vision Transformers Could Be Excellent Teachers (NeurIPS 2024 · similarity 0.86)
- DeepKD: A Deeply Decoupled and Denoised Knowledge Distillation Trainer (NeurIPS 2025 · similarity 0.86)
- Debiased Distillation for Consistency Regularization (AAAI 2025 · similarity 0.86)
- All You Need in Knowledge Distillation Is a Tailored Coordinate System (AAAI 2025 · 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.