We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Towards Adversarial Attack on Vision-Language Pre-training Models

Abstract: While vision-language pre-training model (VLP) has shown revolutionary improvements on various vision-language (V+L) tasks, the studies regarding its adversarial robustness remain largely unexplored. This paper studied the adversarial attack on popular VLP models and V+L tasks. First, we analyzed the performance of adversarial attacks under different settings. By examining the influence of different perturbed objects and attack targets, we concluded some key observations as guidance on both designing strong multimodal adversarial attack and constructing robust VLP models. Second, we proposed a novel multimodal attack method on the VLP models called Collaborative Multimodal Adversarial Attack (Co-Attack), which collectively carries out the attacks on the image modality and the text modality. Experimental results demonstrated that the proposed method achieves improved attack performances on different V+L downstream tasks and VLP models. The analysis observations and novel attack method hopefully provide new understanding into the adversarial robustness of VLP models, so as to contribute their safe and reliable deployment in more real-world scenarios.
Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
Cite as: arXiv:2206.09391 [cs.LG]
  (or arXiv:2206.09391v1 [cs.LG] for this version)

Submission history

From: Jiaming Zhang [view email]
[v1] Sun, 19 Jun 2022 12:55:45 GMT (2958kb,D)

Link back to: arXiv, form interface, contact.