Enregistré dans:
Détails bibliographiques
Auteurs principaux: Guo, Yangyang, Xu, Ziwei, Xu, Xilie, Wong, YongKang, Nie, Liqiang, Kankanhalli, Mohan
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2412.15614
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912163093807104
author Guo, Yangyang
Xu, Ziwei
Xu, Xilie
Wong, YongKang
Nie, Liqiang
Kankanhalli, Mohan
author_facet Guo, Yangyang
Xu, Ziwei
Xu, Xilie
Wong, YongKang
Nie, Liqiang
Kankanhalli, Mohan
contents This technical report introduces our top-ranked solution that employs two approaches, \ie suffix injection and projected gradient descent (PGD) , to address the TiFA workshop MLLM attack challenge. Specifically, we first append the text from an incorrectly labeled option (pseudo-labeled) to the original query as a suffix. Using this modified query, our second approach applies the PGD method to add imperceptible perturbations to the image. Combining these two techniques enables successful attacks on the LLaVA 1.5 model.
format Preprint
id arxiv_https___arxiv_org_abs_2412_15614
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Technical Report for ICML 2024 TiFA Workshop MLLM Attack Challenge: Suffix Injection and Projected Gradient Descent Can Easily Fool An MLLM
Guo, Yangyang
Xu, Ziwei
Xu, Xilie
Wong, YongKang
Nie, Liqiang
Kankanhalli, Mohan
Cryptography and Security
Computer Vision and Pattern Recognition
This technical report introduces our top-ranked solution that employs two approaches, \ie suffix injection and projected gradient descent (PGD) , to address the TiFA workshop MLLM attack challenge. Specifically, we first append the text from an incorrectly labeled option (pseudo-labeled) to the original query as a suffix. Using this modified query, our second approach applies the PGD method to add imperceptible perturbations to the image. Combining these two techniques enables successful attacks on the LLaVA 1.5 model.
title Technical Report for ICML 2024 TiFA Workshop MLLM Attack Challenge: Suffix Injection and Projected Gradient Descent Can Easily Fool An MLLM
topic Cryptography and Security
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.15614