Metraipolis › archive › dc_cf501b67413b

Difficulty Calibration Across Time for Membership Inference in Diffusion Models — arXiv:2610.40616

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.40616
License (as archived)
CC0-1.0
Archived
2026-11-01T02:21:26.751Z
SHA-256 of the archived text
9b8e75a7b9f0a44b856dc4b0cb4ad63b3f1ae8225f0136fa6424ec83d1acefa8
Stable URL of this page
https://metraipolis.com/documento/dc_cf501b67413b

How to cite

Difficulty Calibration Across Time for Membership Inference in Diffusion Models — arXiv:2610.40616 — arXiv (en), https://arxiv.org/abs/2610.40616 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cf501b67413b (Metraipolis archive dc_cf501b67413b, sha256 9b8e75a7b9f0a44b856dc4b0cb4ad63b3f1ae8225f0136fa6424ec83d1acefa8)
BibTeX
@misc{dccf501b67413b,
  title = {Difficulty Calibration Across Time for Membership Inference in Diffusion Models — arXiv:2610.40616},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40616},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cf501b67413b; sha256 9b8e75a7b9f0a44b856dc4b0cb4ad63b3f1ae8225f0136fa6424ec83d1acefa8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40616. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Modern diffusion models are trained on massive datasets that often include copyrighted or private images. Membership inference attacks (MIAs) address this by testing whether a given image was in a model's training set. The standard approach scores an image by its reconstruction loss and flags low-loss images as members. This often fails because some images are simply easier to reconstruct than others, so a simple unseen image can score a lower loss than a complex training image. Difficulty calibration separates memorization from how difficult an image is to begin with. We show how to calibrate along two axes of time: training time and diffusion time. First, across training time, we introduce Reference-Model Calibration (RMC), which compares a fine-tuned model's loss against an earlier checkpoint. This cleanly isolates newly memorized data and gives near-perfect auditing, even without text captions. Since earlier checkpoints are not always available, we next present Temporal Self-Calibration (TSC), which calibrates across diffusion time using only a single model. Our key observation is that an image's difficulty persists across noise levels, whereas memorization is localized in time. By comparing the loss at a target timestep against neighboring timesteps from the same model, TSC attenuates image difficulty using the model's own trajectory. Empirically, our methods beat the state of the art by a wide margin, and unlike most existing attacks, generalize across different model parameterizations such as $x$-, $v$-, and $\epsilon$-prediction models. Code is available at this https URL .

Tema degli esploratori: fiducia digitale e crittografia (elenco arXiv cs.CR, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Cryptography and Security (cs.CR); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.40616 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-01T02:12:13.424Z dalla pagina https://arxiv.org/list/cs.CR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).