Research map / Adversarial robustness and security / Attacks on text and graphs
Content-based Unrestricted Adversarial Attack
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 |
|---|---|---|---|
| add momentum to adversarial latent optimization (ablation) | ACA · ImageNet-compatible, ViT-B surrogate | transfer ASR | Helped |
| add differentiable boundary processing on top of momentum (ablation) | ACA · ImageNet-compatible, ViT-B surrogate | transfer ASR | Helped |
| add Image Latent Mapping (diffusion inversion) (ablation) | ACA · ImageNet-compatible | ASR / image quality | No clear effect |
| generate unrestricted adversarial examples via diffusion latents instead of color-based NCF | ACA vs NCF · ImageNet-compatible, ViT-B surrogate | Avg. black-box ASR (%): 35.39 -> 58.68 | Helped |
| use ACA instead of NCF against defense models | ACA vs NCF · Defense models | Avg. ASR (%): 31.18 -> 58.05 | Helped |
| use ACA instead of ColorFool for image quality | ACA vs ColorFool · Image quality | NIMA-AVA: 5.24 -> 5.54 | Helped |
| use diffusion-based ACA instead of NCF | ACA vs NCF · Attack time (MN-v2) | time (sec): 10.45 -> 125.33 | Hurt |
| use ACA instead of NCF on MN-v2 surrogate for white-box-like ResNet targets | ACA vs NCF · Surrogate RN-50, RN-152 target | ASR | No clear effect |
Closest papers by meaning
- Unrestricted Adversarial Examples via Semantic Manipulation (ICLR 2020 · similarity 0.89)
- Beyond ImageNet Attack: Towards Crafting Adversarial Examples for Black-box Domains (ICLR 2022 · similarity 0.87)
- ObjectAdv: Object-Level Unrestricted Adversarial Attacks via Diffusion Models (AAAI 2026 · similarity 0.87)
- DiffAdvMAP: Flexible Diffusion-Based Framework for Generating Natural Unrestricted Adversarial Examples (ICML 2025 · similarity 0.86)
- Chain of Attack: On the Robustness of Vision-Language Models Against Transfer-Based Adversarial Attacks (CVPR 2025 · similarity 0.86)
- AdvI2I: Adversarial Image Attack on Image-to-Image Diffusion Models (ICML 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.