Metraipolis › archive › dc_bda6e48e97a5
Calibrating One-Round Membership Inference with Neighbors — arXiv:2609.36331
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.36331
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T16:25:38.793Z
- SHA-256 of the archived text
9fc51bc6b129dcd781d443f80ca8b7ccfd48ea77aea8fb99160afc7b76d20266- Stable URL of this page
https://metraipolis.com/documento/dc_bda6e48e97a5
How to cite
Calibrating One-Round Membership Inference with Neighbors — arXiv:2609.36331 — arXiv (en), https://arxiv.org/abs/2609.36331 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bda6e48e97a5 (Metraipolis archive dc_bda6e48e97a5, sha256 9fc51bc6b129dcd781d443f80ca8b7ccfd48ea77aea8fb99160afc7b76d20266)
BibTeX
@misc{dcbda6e48e97a5,
title = {Calibrating One-Round Membership Inference with Neighbors — arXiv:2609.36331},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.36331},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bda6e48e97a5; sha256 9fc51bc6b129dcd781d443f80ca8b7ccfd48ea77aea8fb99160afc7b76d20266},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36331. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
The state-of-the-art Membership Inference (MI) methods calibrate their signal separately for each example using reference models, auxiliary models trained to exclude the target. This paradigm scales poorly to modern large models, however, whose training is too expensive to replicate. This has motivated one-round settings, where only a single trained model is available; but without reference models the per-example calibration that drives the strongest attacks can no longer be estimated, leaving the membership signal weak. We ask whether neighbors of the target point can recover this calibration without training any additional model. Our key observation is that reference models serve only to reveal how an example behaves under models not trained on it, and that querying the target model on nearby samples yields the same information. We propose two complementary ways to obtain such neighbors, and show that querying them against an early training checkpoint further sharpens the signal. We evaluate across three image classification datasets and three training setups, showing that neighbors yield strong membership signals and competitive attack performance at no additional training cost. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.AI). Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.36331 — 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 2026-10-07T15:54:53.812Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).