Research map / Adversarial robustness and security / Attacks on text and graphs
One Perturbation is Enough: On Generating Universal Adversarial Perturbations against Vision-Language Pre-training Models
Hao Fang, Jiawei Kong, Wenbo Yu, Bin Chen, Jiawei Li, Hao Wu, Shu-Tao Xia, Ke Xu
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 contrastive loss L_CL (ablation) | C-PGC · Flickr30K TR, ALBEF to TCL | ASR (%): 34.99 -> 62.11 | Improved |
| add unimodal distance loss L_Dis (ablation) | C-PGC · Flickr30K TR, ALBEF white-box | ASR (%): 79.54 -> 90.13 | Improved |
| use farthest selection instead of random positives (ablation) | C-PGC · Flickr30K, ALBEF white-box TR | ASR (%): 61.87 -> 90.13 | Improved |
| add cross-attention cross-modal conditioning to generator (ablation) | C-PGC · Flickr30K TR, ALBEF to TCL | ASR (%): 45.76 -> 62.11 | Improved |
| L_CL with random positives instead of none (ablation) | C-PGC · Flickr30K TR, ALBEF white-box | ASR (%): 76.46 -> 61.87 | Hurt |
| use generative contrastive C-PGC instead of ETU (ablation) | ETU · Flickr30K, black-box | ASR (%): +17.76% | Improved |
| set lambda to 0.1 (ablation) | C-PGC · Flickr30K | ASR (%) | Improved |
| increase image budget from 12/255 to 16/255 (ablation) | C-PGC · Flickr30K | ASR (%) | No clear effect |
| reduce image budget to 4/255 (ablation) | C-PGC · Flickr30K | ASR (%) | Hurt |
Closest papers by meaning
- Pandora's Box: Towards Building Universal Attackers against Real-World Large Vision-Language Models (NeurIPS 2024 · similarity 0.90)
- ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time (ICLR 2025 · similarity 0.89)
- One Prompt Word is Enough to Boost Adversarial Robustness for Pre-trained Vision-Language Models (CVPR 2024 · similarity 0.88)
- SLADE: Shielding against Dual Exploits in Large Vision-Language Models (CVPR 2025 · similarity 0.88)
- Attention! Your Vision Language Model Could Be Maliciously Manipulated (NeurIPS 2025 · similarity 0.88)
- HQA-VLAttack: Towards High Quality Adversarial Attack on Vision-Language Pre-Trained Models (NeurIPS 2025 · similarity 0.88)
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.
