Research map / Adversarial robustness and security / Attacks on text and graphs
Discrete Point-Wise Attack Is Not Enough: Generalized Manifold Adversarial Attack for Face Recognition
Read the paperCheck 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 |
|---|---|---|---|
| Train on the target identity state set rather than one target image | GMAA · Target 1 | ASR (Facenet): 4.54 -> 17.84 | Helped |
| Train on the target identity state set rather than one target image | GMAA · Target 1 | ASR (Mobileface): 43.44 -> 67.50 | Helped |
| Train on the target identity state set rather than one target image | GMAA · Target 2 | ASR (Facenet): 13.30 -> 21.71 | Helped |
| Train on the target identity state set rather than one target image | GMAA · Target 2 | ASR (Mobileface): 33.08 -> 41.24 | Helped |
| Train on the target identity state set rather than one target image | GMAA · Target 3 | ASR (Facenet): 12.64 -> 19.15 | Helped |
| Train on the target identity state set rather than one target image | GMAA · Target 3 | ASR (Mobileface): 29.56 -> 47.21 | Helped |
| Use a state set with more than one image for each expression state (ablation) | GMAA and generalized baselines · Targets 1–3 | ASR: 1/ 2/ 3 image(s) | No clear effect |
| Add AU predictor D_AU (ablation) | GMAA generator · Generated adversarial examples | AU-vector MSE: 0.5549 -> 0.3582 | Helped |
| Add three local editors (ablation) | GMAA generator · Generated adversarial examples | AU-vector MSE: 0.6283 -> 0.3582 | Helped |
| Add three local editors (ablation) | GMAA generator · Generated adversarial examples | visual quality | Helped |
| Expand adversarial generation from discrete examples to an AU-conditioned manifold | MAA · CelebA-HQ and LFW | black-box ASR | Mixed |
Closest papers by meaning
- Dual Manifold Adversarial Robustness: Defense against Lp and non-Lp Adversarial Attacks (NeurIPS 2020 · similarity 0.87)
- Non-Adaptive Adversarial Face Generation (NeurIPS 2025 · similarity 0.87)
- Manifold Constraints for Imperceptible Adversarial Attacks on Point Clouds (AAAI 2024 · similarity 0.84)
- Improving the Transferability of Adversarial Attacks on Face Recognition with Diverse Parameters Augmentation (CVPR 2025 · similarity 0.84)
- Enhancing the Adversarial Robustness via Manifold Projection (AAAI 2025 · similarity 0.84)
- Adv-Diffusion: Imperceptible Adversarial Face Identity Attack via Latent Diffusion Model (AAAI 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.