Research map / Adversarial robustness and security / Backdoor and poisoning
Black-box Backdoor Defense via Zero-shot Image Purification
Yucheng Shi, Mengnan Du, Xuansheng Wu, Zihan Guan, Jin Sun, Ninghao Liu
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 |
|---|---|---|---|
| Apply ZIP purification | ResNet-34 poisoned classifier · Imagenette BadNet | ASR (%): 94.53 -> 7.55 | Improved |
| Apply ZIP purification | ResNet-34 poisoned classifier · GTSRB PhysicalBA | ASR (%): 100.00 -> 6.57 | Improved |
| Use ZIP instead of ShrinkPad | Black-box defense · CIFAR-10 PhysicalBA | ASR (%): 90.50 -> 4.33 | Improved |
| Add diffusion recovery after blur (ZIP vs Blur only) | Purification · CIFAR-10 average | clean accuracy (%): 51.98 -> 77.23 | Improved |
| Use ShrinkPad defense (no-defense baseline) | Black-box defense · CIFAR-10 BadNet | clean accuracy (%): 82.31 -> 62.89 | Hurt |
| Use stronger blur (Blur2 to Blur8) (ablation) | ZIP · Imagenette Blended | ASR (%): 26.11 -> 7.69 | Improved |
| Use stronger blur (Blur2 to Blur8) (ablation) | ZIP · Imagenette Blended | clean accuracy (%): 84.05 -> 71.03 | Hurt |
| Replace Blur4 with Grayscale transformation (ablation) | ZIP · Imagenette Blended | ASR (%): 8.35 -> 99.40 | Hurt |
| Use ZIP instead of BDMAE masking and reconstruction | Purification · CIFAR-10 Blended | ASR (%): 99.88 -> 7.75 | Improved |
| Attacker trains with purified images using same transformation (Blur+Grayscale) (ablation) | ZIP · Enhanced BadNets | ASR (%): 24.00 -> 79.89 | Hurt |
| Defend enhanced attack with Blur only instead of Blur+Grayscale (ablation) | ZIP · Enhanced BadNets | ASR (%): 79.89 -> 18.47 | Improved |
| Use ZIP instead of DiffPure | Purification · Imagenette Blended | ASR (%): 43.82 -> 8.35 | Improved |
Closest papers by meaning
- UFID: A Unified Framework for Black-box Input-level Backdoor Detection on Diffusion Models (AAAI 2025 · similarity 0.90)
- SampDetox: Black-box Backdoor Defense via Perturbation-based Sample Detoxification (NeurIPS 2024 · similarity 0.89)
- TERD: A Unified Framework for Safeguarding Diffusion Models Against Backdoors (ICML 2024 · similarity 0.88)
- Elijah: Eliminating Backdoors Injected in Diffusion Models via Distribution Shift (AAAI 2024 · similarity 0.88)
- Clean Data Can Still Carry Backdoors: Support-Persistent Backdoors for Model Reuse (NeurIPS 2026 · similarity 0.88)
- Uncovering, Explaining, and Mitigating the Superficial Safety of Backdoor Defense (NeurIPS 2024 · similarity 0.87)
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.
