Saved in:
Bibliographic Details
Main Authors: Peng, Duo, Ke, Qiuhong, Liu, Jun
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.11336
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913363806650368
author Peng, Duo
Ke, Qiuhong
Liu, Jun
author_facet Peng, Duo
Ke, Qiuhong
Liu, Jun
contents Text-to-Image (T2I) models have raised security concerns due to their potential to generate inappropriate or harmful images. In this paper, we propose UPAM, a novel framework that investigates the robustness of T2I models from the attack perspective. Unlike most existing attack methods that focus on deceiving textual defenses, UPAM aims to deceive both textual and visual defenses in T2I models. UPAM enables gradient-based optimization, offering greater effectiveness and efficiency than previous methods. Given that T2I models might not return results due to defense mechanisms, we introduce a Sphere-Probing Learning (SPL) scheme to support gradient optimization even when no results are returned. Additionally, we devise a Semantic-Enhancing Learning (SEL) scheme to finetune UPAM for generating target-aligned images. Our framework also ensures attack stealthiness. Extensive experiments demonstrate UPAM's effectiveness and efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2405_11336
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle UPAM: Unified Prompt Attack in Text-to-Image Generation Models Against Both Textual Filters and Visual Checkers
Peng, Duo
Ke, Qiuhong
Liu, Jun
Computer Vision and Pattern Recognition
I.2.6
Text-to-Image (T2I) models have raised security concerns due to their potential to generate inappropriate or harmful images. In this paper, we propose UPAM, a novel framework that investigates the robustness of T2I models from the attack perspective. Unlike most existing attack methods that focus on deceiving textual defenses, UPAM aims to deceive both textual and visual defenses in T2I models. UPAM enables gradient-based optimization, offering greater effectiveness and efficiency than previous methods. Given that T2I models might not return results due to defense mechanisms, we introduce a Sphere-Probing Learning (SPL) scheme to support gradient optimization even when no results are returned. Additionally, we devise a Semantic-Enhancing Learning (SEL) scheme to finetune UPAM for generating target-aligned images. Our framework also ensures attack stealthiness. Extensive experiments demonstrate UPAM's effectiveness and efficiency.
title UPAM: Unified Prompt Attack in Text-to-Image Generation Models Against Both Textual Filters and Visual Checkers
topic Computer Vision and Pattern Recognition
I.2.6
url https://arxiv.org/abs/2405.11336