Membership Inference Attacks for Face Images Against Fine-Tuned Latent Diffusion Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910858577182720 |
|---|---|
| author | Holme, Lauritz Christian Storgaard, Anton Mosquera Bigdeli, Siavash Arjomand |
| author_facet | Holme, Lauritz Christian Storgaard, Anton Mosquera Bigdeli, Siavash Arjomand |
| contents | The rise of generative image models leads to privacy concerns when it comes to the huge datasets used to train such models. This paper investigates the possibility of inferring if a set of face images was used for fine-tuning a Latent Diffusion Model (LDM). A Membership Inference Attack (MIA) method is presented for this task. Using generated auxiliary data for the training of the attack model leads to significantly better performance, and so does the use of watermarks. The guidance scale used for inference was found to have a significant influence. If a LDM is fine-tuned for long enough, the text prompt used for inference has no significant influence. The proposed MIA is found to be viable in a realistic black-box setup against LDMs fine-tuned on face-images. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_11619 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Membership Inference Attacks for Face Images Against Fine-Tuned Latent Diffusion Models Holme, Lauritz Christian Storgaard, Anton Mosquera Bigdeli, Siavash Arjomand Computer Vision and Pattern Recognition The rise of generative image models leads to privacy concerns when it comes to the huge datasets used to train such models. This paper investigates the possibility of inferring if a set of face images was used for fine-tuning a Latent Diffusion Model (LDM). A Membership Inference Attack (MIA) method is presented for this task. Using generated auxiliary data for the training of the attack model leads to significantly better performance, and so does the use of watermarks. The guidance scale used for inference was found to have a significant influence. If a LDM is fine-tuned for long enough, the text prompt used for inference has no significant influence. The proposed MIA is found to be viable in a realistic black-box setup against LDMs fine-tuned on face-images. |
| title | Membership Inference Attacks for Face Images Against Fine-Tuned Latent Diffusion Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2502.11619 |