作者位次 #3/6ACL 2026Findings· Findings of ACL
Asymmetric Relational-Geometry Driven Universal Adversarial Perturbations for Vision-Language Models
Jiaxin Ye · Weihai Li · Ying Wang · Simeng Qin · Zhitao Zeng · Zikai Xu
摘要
Although vision-language pre-trained (VLP) models have achieved remarkable success across multimodal tasks, they remain vulnerable to adversarial perturbations.Existing universal adversarial perturbation (UAP) methods in multimodal settings—whether generator-based or optimization-based—often suff...