Saved in:
Bibliographic Details
Main Authors: Xu, Xide, Butt, Muhammad Atif, Kamath, Sandesh, Raducanu, Bogdan
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.16437
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913585262755840
author Xu, Xide
Butt, Muhammad Atif
Kamath, Sandesh
Raducanu, Bogdan
author_facet Xu, Xide
Butt, Muhammad Atif
Kamath, Sandesh
Raducanu, Bogdan
contents The growing demand for customized visual content has led to the rise of personalized text-to-image (T2I) diffusion models. Despite their remarkable potential, they pose significant privacy risk when misused for malicious purposes. In this paper, we propose a novel and efficient adversarial attack method, Concept Protection by Selective Attention Manipulation (CoPSAM) which targets only the cross-attention layers of a T2I diffusion model. For this purpose, we carefully construct an imperceptible noise to be added to clean samples to get their adversarial counterparts. This is obtained during the fine-tuning process by maximizing the discrepancy between the corresponding cross-attention maps of the user-specific token and the class-specific token, respectively. Experimental validation on a subset of CelebA-HQ face images dataset demonstrates that our approach outperforms existing methods. Besides this, our method presents two important advantages derived from the qualitative evaluation: (i) we obtain better protection results for lower noise levels than our competitors; and (ii) we protect the content from unauthorized use thereby protecting the individual's identity from potential misuse.
format Preprint
id arxiv_https___arxiv_org_abs_2411_16437
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Privacy Protection in Personalized Diffusion Models via Targeted Cross-Attention Adversarial Attack
Xu, Xide
Butt, Muhammad Atif
Kamath, Sandesh
Raducanu, Bogdan
Computer Vision and Pattern Recognition
Machine Learning
The growing demand for customized visual content has led to the rise of personalized text-to-image (T2I) diffusion models. Despite their remarkable potential, they pose significant privacy risk when misused for malicious purposes. In this paper, we propose a novel and efficient adversarial attack method, Concept Protection by Selective Attention Manipulation (CoPSAM) which targets only the cross-attention layers of a T2I diffusion model. For this purpose, we carefully construct an imperceptible noise to be added to clean samples to get their adversarial counterparts. This is obtained during the fine-tuning process by maximizing the discrepancy between the corresponding cross-attention maps of the user-specific token and the class-specific token, respectively. Experimental validation on a subset of CelebA-HQ face images dataset demonstrates that our approach outperforms existing methods. Besides this, our method presents two important advantages derived from the qualitative evaluation: (i) we obtain better protection results for lower noise levels than our competitors; and (ii) we protect the content from unauthorized use thereby protecting the individual's identity from potential misuse.
title Privacy Protection in Personalized Diffusion Models via Targeted Cross-Attention Adversarial Attack
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2411.16437