AI Paper Atlas

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

PA-Attack: Guiding Gray-Box Attacks on LVLM Vision Encoders with Prototypes and Attention

Hefei Mei, Zirui Wang, Chang Xu, Jianyuan Guo, Minjing Dong

CVPR 2026 · Adversarial robustness and security › Black-box and transfer attacks · approach: Transfer attacks on face recognition · 2 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
PA-Attack vs VEAttack at eps=2/255gray-box LVLM attack · LLaVA-1.5-7B average SRRaverage SRR %: 65.2 -> 77.1Improved
PA-Attack vs VEAttack on OpenFlamingo-9B at eps=4/255gray-box LVLM attack · OpenFlamingo-9B average SRRaverage SRR %: 61.5 -> 66.4Improved
add prototype-anchored guidance (PG) (ablation)vision encoder attack (100 steps) · LLaVA-1.5-7B TextVQASRR %: 72.8 -> 76.3Improved
add attention enhancement (AE) alone (ablation)vision encoder attack (100 steps) · LLaVA-1.5-7B, COCO/Flickr30k/TextVQA/VQAv2SRR %: VQAv2 48.4 -> 50.9; TextVQA 72.8 -> 72.3Mixed
add AE on top of PG (ablation)PA-Attack (100 steps) · LLaVA-1.5-7B VQAv2SRR %: 52.2 -> 54.2Improved
add two-stage attention refinement (TS) (ablation)PA-Attack (PG+AE, 150 steps) · LLaVA-1.5-7B TextVQASRR %: 81.1 -> 86.3Improved
use nearest guidance sample instead of farthest prototype (ablation)PA-Attack · LLaVA-1.5-7B guidance ablationaverage SRR %: 82.9 -> 61.4Hurt
use farthest prototype instead of farthest sample (ablation)PA-Attack · LLaVA-1.5-7B guidance ablationaverage SRR %: 80.6 -> 82.9Improved
increase loss weight lambda from 1.0 to 2.0 (ablation)PA-Attack · LLaVA-1.5-7Baverage SRR %: 83.7 -> 82.5Hurt
decrease lambda from 1.0 to 0.5 (ablation)PA-Attack · LLaVA-1.5-7Baverage SRR %: 83.7 -> 81.7Hurt
increase PCA dimension from 1024 to 2048 (ablation)PA-Attack prototypes · LLaVA-1.5-7Baverage SRR % (Flickr30k, TextVQA, VQAv2): 74.4 -> 70.7Hurt
change number of prototypes K from 20 to 10 or 30 (ablation)PA-Attack prototypes · LLaVA-1.5-7Baverage SRR %: 74.4 vs 74.3 (K=10) vs 73.6 (K=30)No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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