AI Paper Atlas

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

TPAMI 2025 · Adversarial robustness and security › Attacks on text and graphs · approach: Vision-language model robustness · 18 citations (OpenAlex)

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 changedWhereMeasuredEffect
use SA-AET instead of DRAadversarial attack · Flickr30K ALBEF->CLIP-CNNTR R@1 ASR: 49.55 -> 57.22Improved
use SA-AET instead of DRAadversarial attack · Flickr30K ALBEF->CLIP-CNNIR R@1 ASR: 59.01 -> 65.59Improved
use SA-AET instead of SGAadversarial attack · Flickr30K ALBEF->TCLTR R@1 ASR: 87.88 -> 96.42Improved
use SA-AET instead of DRAadversarial attack · MSCOCO CLIP-ViT->TCLTR R@1 ASR: 51.88 -> 57.30Improved
use SA-AET instead of DRAadversarial attack · MSCOCO CLIP-ViT->TCLIR R@1 ASR: 61.06 -> 65.16Improved
use SA-AET instead of DRAadversarial attack · Image captioning transferCIDEr: 88.3 -> 65.7Improved
use SA-AET instead of DRAadversarial attack · GPT-4 transferR@1 ASR: 16.0 -> 18.0Improved
use SA-AET instead of DRAadversarial attack · Qwen-VL transferR@1 ASR: 15.0 -> 16.0Improved
increase number of samples m beyond 5 from adversarial trajectory (ablation)SA-AET · Flickr30K from ALBEFASRNo clear effect
use 40% of text data for projection matrix instead of none (DRA baseline) (ablation)SA-AET · Flickr30K ALBEF->CLIP-ViT TRASR: 46.26 -> 49.08Improved
vary proportion of text data used for projection matrix (10% to 100%) (ablation)SA-AET · Flickr30K ALBEF->TCLASR: 94.42 to 95.15No clear effect

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy