AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

DualTAP: A Dual-Task Adversarial Protector for Mobile MLLM Agents

Fuyao Zhang, Jiaming Zhang, CHE WANG, Xiongtao Sun, Yurong Hao, Guowei Guan, Wenjie Li, Longtao Huang, Wei Lim

ECCV 2026 · Multi-agent systems and games › LLM agents · approach: GUI agent memory and trajectories · 10 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 contrastive attention to blur perturbation (ablation)Blur · PrivScreennormal-task accuracy: 33% -> 54%Improved
use dual-task objective with attention instead of mask+attention (ablation)DualTAP · PrivScreennormal-task accuracy: 52% -> over 80%Improved
increase perturbation bound epsilon from 64/255 to 192/255 (ablation)DualTAP · PrivScreentask accuracy: 88% -> 72%Hurt
increase perturbation bound epsilon from 64/255 to 192/255 (ablation)DualTAP · PrivScreenprivacy metrics: 80~100% -> 20~40%Improved
use DualTAP instead of FOA-AttackQwen2.5-VL-7B · PrivScreenleakage rate: 83.81 -> 32.86Improved
use DualTAP instead of no protectionGPT-5 · PrivScreenBERTScore: 0.9260 -> 0.1589Improved
use DualTAP instead of no protectionUI-TARS-7B · PrivScreentask accuracy: 80.00 -> 67.00Hurt
use DualTAP instead of no protectionInternVL3-5-8B · PrivScreentask accuracy: 83.00 -> 83.00No 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