AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

PLA: Prompt Learning Attack against Text-to-Image Generative Models

Xinqi Lyu, Yihao Liu, Yanjie Li, Bin Xiao

ICCV 2025 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 3 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
add text-image loss L_a (ablation)PLA-T5 on SLD · Violence, SCASR-4 (%): 81.02 -> 93.34Improved
add image-image loss L_b (ablation)PLA-T5 on SLD · Violence, SCASR-4 (%): 79.34 -> 93.34Improved
add text-image loss L_a (ablation)PLA-T5 on SLD · Nudity, SCASR-4 (%): 82.99 -> 93.41Improved
add image-image loss L_b (ablation)PLA-T5 on SLD · Nudity, SCASR-4 (%): 74.66 -> 93.41Improved
replace ZOO gradient with history-retaining gradient (ablation)PLA-BERT on SDXLv1.0 · Violence, SCASR-4 (%): 52.89 -> 91.69Improved
replace ZOO gradient with history-retaining gradient (ablation)PLA-BERT on SDXLv1.0 · Nudity, SCASR-4 (%): 58.44 -> 95.37Improved
add restart strategy (Gaussian noise replaces black images) (ablation)PLA-BERT on SDXLv1.0 · Violence, SCASR-4 (%): 70.12 -> 91.69Improved
add restart strategy (Gaussian noise replaces black images) (ablation)PLA-BERT on SDXLv1.0 · Nudity, SCASR-4 (%): 78.33 -> 95.37Improved
use T5 instead of BERT as the language modelPLA · Nudity, avg ASR-4 on SDv1.5ASR-4 (%): 91.45 -> 86.56Hurt
use T5 instead of BERT as the language modelPLA · Violence, avg ASR-4 on SDv1.5ASR-4 (%): 88.62 -> 90.41Improved
use PLA instead of MMA-Diffusionattack method · Nudity, avg ASR-4 on SDXLv1.0ASR-4 (%): 73.30 -> 90.57Improved
use PLA-T5 instead of MMA-Diffusion on online serviceattack method · Violence, Stability.aiASR-4 (%): 42.57 -> 69.70Improved

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