Saved in:
Bibliographic Details
Main Authors: Liu, Xiao, Liu, Jiaxiang, Peng, Boci, Hu, Boren, Wang, Yusong, Chen, Xiwen, Tiwari, Prayag, Zhang, Liming, Xu, Mingkun
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.25922
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916044860293120
author Liu, Xiao
Liu, Jiaxiang
Peng, Boci
Hu, Boren
Wang, Yusong
Chen, Xiwen
Tiwari, Prayag
Zhang, Liming
Xu, Mingkun
author_facet Liu, Xiao
Liu, Jiaxiang
Peng, Boci
Hu, Boren
Wang, Yusong
Chen, Xiwen
Tiwari, Prayag
Zhang, Liming
Xu, Mingkun
contents Vision Language Models adapt well to downstream tasks but are highly vulnerable to adversarial perturbations that disrupt cross-modal semantic alignment. Existing defenses are largely unidirectional or structural, failing to exploit bidirectional cross-modal complementarity and instance-wise adaptive protection. To overcome the limitations of unidirectional and static defenses in adversarial settings, we propose Closed-Loop Bidirectional Prompting, casting robust adaptation as cross-modal agreement recovery via a dynamic feedback loop on frozen encoders. A Semantic Anchor is introduced as a stable prior to constrain cyclic updates and mitigate perturbation-induced feature corruption. Through anchor-based bootstrapping, textual semantics denoise visual representations, while the refined visuals enable instance-adaptive prompt updating, yielding a rectified and robust consensus. Extensive evaluations across 11 datasets validate state-of-the-art robustness and strong base-to-new generalization, while maintaining a favorable trade-off between computational cost and accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2605_25922
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Closed-Loop Bidirectional Prompting for Adversarial Robustness of Vision Language Models
Liu, Xiao
Liu, Jiaxiang
Peng, Boci
Hu, Boren
Wang, Yusong
Chen, Xiwen
Tiwari, Prayag
Zhang, Liming
Xu, Mingkun
Computer Vision and Pattern Recognition
Vision Language Models adapt well to downstream tasks but are highly vulnerable to adversarial perturbations that disrupt cross-modal semantic alignment. Existing defenses are largely unidirectional or structural, failing to exploit bidirectional cross-modal complementarity and instance-wise adaptive protection. To overcome the limitations of unidirectional and static defenses in adversarial settings, we propose Closed-Loop Bidirectional Prompting, casting robust adaptation as cross-modal agreement recovery via a dynamic feedback loop on frozen encoders. A Semantic Anchor is introduced as a stable prior to constrain cyclic updates and mitigate perturbation-induced feature corruption. Through anchor-based bootstrapping, textual semantics denoise visual representations, while the refined visuals enable instance-adaptive prompt updating, yielding a rectified and robust consensus. Extensive evaluations across 11 datasets validate state-of-the-art robustness and strong base-to-new generalization, while maintaining a favorable trade-off between computational cost and accuracy.
title Closed-Loop Bidirectional Prompting for Adversarial Robustness of Vision Language Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.25922