Research map / Adversarial robustness and security / Attacks on text and graphs
Semantic-Aligned Adversarial Evolution Triangle for High-Transferability Vision-Language Attack
Xiaojun Jia, Sensen Gao, Qing Guo, Simeng Qin, Ke Ma, Yihao Huang, Yang Liu, Ivor Wai-Hung Tsang, Xiaochun Cao
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 SA-AET instead of DRA | adversarial attack · Flickr30K ALBEF->CLIP-CNN | TR R@1 ASR: 49.55 -> 57.22 | Improved |
| use SA-AET instead of DRA | adversarial attack · Flickr30K ALBEF->CLIP-CNN | IR R@1 ASR: 59.01 -> 65.59 | Improved |
| use SA-AET instead of SGA | adversarial attack · Flickr30K ALBEF->TCL | TR R@1 ASR: 87.88 -> 96.42 | Improved |
| use SA-AET instead of DRA | adversarial attack · MSCOCO CLIP-ViT->TCL | TR R@1 ASR: 51.88 -> 57.30 | Improved |
| use SA-AET instead of DRA | adversarial attack · MSCOCO CLIP-ViT->TCL | IR R@1 ASR: 61.06 -> 65.16 | Improved |
| use SA-AET instead of DRA | adversarial attack · Image captioning transfer | CIDEr: 88.3 -> 65.7 | Improved |
| use SA-AET instead of DRA | adversarial attack · GPT-4 transfer | R@1 ASR: 16.0 -> 18.0 | Improved |
| use SA-AET instead of DRA | adversarial attack · Qwen-VL transfer | R@1 ASR: 15.0 -> 16.0 | Improved |
| increase number of samples m beyond 5 from adversarial trajectory (ablation) | SA-AET · Flickr30K from ALBEF | ASR | No clear effect |
| use 40% of text data for projection matrix instead of none (DRA baseline) (ablation) | SA-AET · Flickr30K ALBEF->CLIP-ViT TR | ASR: 46.26 -> 49.08 | Improved |
| vary proportion of text data used for projection matrix (10% to 100%) (ablation) | SA-AET · Flickr30K ALBEF->TCL | ASR: 94.42 to 95.15 | No clear effect |
Closest papers by meaning
- Boosting Transferability in Vision-Language Attacks via Diversification along the Intersection Region of Adversarial Trajectory (ECCV 2024 · similarity 0.93)
- Chain of Attack: On the Robustness of Vision-Language Models Against Transfer-Based Adversarial Attacks (CVPR 2025 · similarity 0.89)
- Visual Adversarial Examples Jailbreak Aligned Large Language Models (AAAI 2024 · similarity 0.88)
- Failures to Find Transferable Image Jailbreaks Between Vision-Language Models (ICLR 2025 · similarity 0.87)
- A Frustratingly Simple Yet Highly Effective Attack Baseline: Over 90% Success Rate Against the Strong Black-box Models of GPT-4.5/4o/o1 (NeurIPS 2025 · similarity 0.87)
- Harnessing the Computation Redundancy in ViTs to Boost Adversarial Transferability (NeurIPS 2025 · 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.
