Research map / Adversarial robustness and security / Attacks on text and graphs
AdversariaLeak: External Information Leakage Attack Using Adversarial Samples on Face Recognition Systems
Roye Katzav, Amit Giloni, Edita Grolman, Hiroo Saito, Tomoyuki Shibata, Tsukasa Omino, Misaki Komatsu, Yoshikazu Hanatani, Yuval Elovici, Asaf Shabtai
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 |
|---|---|---|---|
| use AdversariaLeak (unique adversarial samples) instead of loss test attack | EIL attack on face verification · CelebA young and male properties | correct property inference | Improved |
| use AdversariaLeak instead of threshold test attack | EIL attack on face verification · CelebA 5 o'clock shadow | correct property inference | Improved |
| use AdversariaLeak instead of loss test attack on 5 o'clock shadow | EIL attack on face verification · CelebA 5 o'clock shadow | correct property inference | No clear effect |
| stricter scenario with different head architecture for target vs substitute | Loss test attack · CelebA | correct property inference | Hurt |
| fine-tune backbone rather than use pre-trained backbone | AdversariaLeak · MAAD-Face vs CelebA, 50% distribution | bias toward specific property value | Improved |
| use unique adversarial samples with small query budget (ablation) | AdversariaLeak · CelebA and MAAD-Face | correct property inference: 24 to 60 queries | Improved |
Closest papers by meaning
- Efficient Decision-Based Black-Box Adversarial Attacks on Face Recognition (CVPR 2019 · similarity 0.86)
- Non-Adaptive Adversarial Face Generation (NeurIPS 2025 · similarity 0.85)
- Transferable Facial Privacy Protection against Blind Face Restoration via Domain-Consistent Adversarial Obfuscation (ICML 2024 · similarity 0.85)
- Unlearn and Burn: Adversarial Machine Unlearning Requests Destroy Model Accuracy (ICLR 2025 · similarity 0.84)
- CryptoFace: End-to-End Encrypted Face Recognition (CVPR 2025 · similarity 0.83)
- BRUSLEATTACK: A QUERY-EFFICIENT SCORE- BASED BLACK-BOX SPARSE ADVERSARIAL ATTACK (ICLR 2024 · similarity 0.83)
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.
