Discovering Universal Semantic Triggers for Text-to-Image Synthesis
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866916123213037568 |
|---|---|
| author | Zhai, Shengfang Wang, Weilong Li, Jiajun Dong, Yinpeng Su, Hang Shen, Qingni |
| author_facet | Zhai, Shengfang Wang, Weilong Li, Jiajun Dong, Yinpeng Su, Hang Shen, Qingni |
| contents | Recently text-to-image models have gained widespread attention in the community due to their controllable and high-quality generation ability. However, the robustness of such models and their potential ethical issues have not been fully explored. In this paper, we introduce Universal Semantic Trigger, a meaningless token sequence that can be added at any location within the input text yet can induce generated images towards a preset semantic target.To thoroughly investigate it, we propose Semantic Gradient-based Search (SGS) framework. SGS automatically discovers the potential universal semantic triggers based on the given semantic targets. Furthermore, we design evaluation metrics to comprehensively evaluate semantic shift of images caused by these triggers. And our empirical analyses reveal that the mainstream open-source text-to-image models are vulnerable to our triggers, which could pose significant ethical threats. Our work contributes to a further understanding of text-to-image synthesis and helps users to automatically auditing their models before deployment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_07562 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Discovering Universal Semantic Triggers for Text-to-Image Synthesis Zhai, Shengfang Wang, Weilong Li, Jiajun Dong, Yinpeng Su, Hang Shen, Qingni Cryptography and Security Artificial Intelligence Recently text-to-image models have gained widespread attention in the community due to their controllable and high-quality generation ability. However, the robustness of such models and their potential ethical issues have not been fully explored. In this paper, we introduce Universal Semantic Trigger, a meaningless token sequence that can be added at any location within the input text yet can induce generated images towards a preset semantic target.To thoroughly investigate it, we propose Semantic Gradient-based Search (SGS) framework. SGS automatically discovers the potential universal semantic triggers based on the given semantic targets. Furthermore, we design evaluation metrics to comprehensively evaluate semantic shift of images caused by these triggers. And our empirical analyses reveal that the mainstream open-source text-to-image models are vulnerable to our triggers, which could pose significant ethical threats. Our work contributes to a further understanding of text-to-image synthesis and helps users to automatically auditing their models before deployment. |
| title | Discovering Universal Semantic Triggers for Text-to-Image Synthesis |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2402.07562 |