Research map / Adversarial robustness and security / Backdoor and poisoning
Breaking High Confidence: Practical Face Impersonation under High-Security Thresholds
Changjin Kim, Seunghun Paik, Dongsoo Kim, Jae Hong Seo
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 correction matrix (ablation) | NbNet attack · LFW vs AWS CompareFace (threshold 99) | ASR (%): 0.33 -> 2.83 | Improved |
| add PCA-based OFS (ablation) | NbNet attack · LFW vs AWS CompareFace (threshold 99) | ASR (%): 0.33 -> 46.68 | Improved |
| add correction matrix on top of PCA-based OFS (ablation) | NbNet+P · LFW vs AWS CompareFace (threshold 99) | ASR (%): 46.68 -> 56.10 | Improved |
| replace NbNet with SPNet inverse model (ablation) | NbNet+CP · LFW vs AWS CompareFace (threshold 99) | ASR (%): 56.10 -> 92.80 | Improved |
| replace SPNet with Arc2Face high-resolution inverse model (ablation) | SPNet+CP · LFW vs AWS CompareFace (threshold 99) | ASR (%): 92.80 -> 53.92 | Hurt |
| add micro-level modifications (GELU, no tanh, PNorm, tuning) to NbNet (ablation) | NbNet inverse model · LFW inverse model ablation | inversion precision: 0.8809 -> 0.9687 | Improved |
| add micro-level modifications to NbNet (ablation) | NbNet inverse model · LFW vs AWS CompareFace (threshold 99) | ASR (%): 56.10 -> 91.41 ± 1.75 | Improved |
| add DSCasConv on top of micro modifications (ablation) | SPNet · LFW vs AWS CompareFace (threshold 99) | ASR (%): 91.41 ± 1.75 -> 91.88 ± 1.60 | Improved |
| add correction matrix on top of PCA-based OFS (ablation) | NbNet+P · LFW, F_1, FMR 1e-6 | ASR (%): 10.37 -> 11.47 | Improved |
| add correction matrix alone (ablation) | NbNet baseline · LFW, F_2, FMR 1e-6 | ASR (%): 24.13 -> 22.20 | Hurt |
| add PCA-based OFS (ablation) | NbNet baseline · LFW, F_2, FMR 1e-6 | ASR (%): 24.13 -> 69.13 | Improved |
| add PCA-based OFS (ablation) | NbNet baseline · LFW, F_S (white-box), FMR 1e-6 | ASR (%): 16.77 -> 43.47 | Improved |
Closest papers by meaning
- FaceShield: Explainable Face Anti-Spoofing with Multimodal Large Language Models (AAAI 2026 · similarity 0.85)
- Non-Adaptive Adversarial Face Generation (NeurIPS 2025 · similarity 0.84)
- FSFM: A Generalizable Face Security Foundation Model via Self-Supervised Facial Representation Learning (CVPR 2025 · similarity 0.83)
- From Intuition to Investigation: A Tool-Augmented Reasoning MLLM Framework for Generalizable Face Anti-Spoofing (CVPR 2026 · similarity 0.83)
- Test-Time Domain Generalization for Face Anti-Spoofing (CVPR 2024 · similarity 0.82)
- SLIP: Spoof-Aware One-Class Face Anti-Spoofing with Language Image Pretraining (AAAI 2025 · similarity 0.82)
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.
