Newest papers

Privacy Accounting for Empirical Denoisers

William Cheng, Mijung Park, Dongjun Kim · 2026 · arXiv

DOI: 10.48550/arXiv.2610.27710 · open-access PDF

French

Les modèles de diffusion sont apparus comme la base des systèmes génératifs modernes, mais leur capacité de mémorisation élevée soulève des préoccupations en matière de confidentialité. Tandis que la formation à la différence privée (DP) fournit des garanties formelles, elle a un coût en termes de qualité de l'échantillon. Dans ce travail, nous étudions une question complémentaire: combien de perte de confidentialité est subie pendant le processus d'échantillonnage lui-même? L'analyse directe des dénaisseurs neuronaux modernes est difficile car leur sensibilité à un seul exemple de formation est généralement intractable.

Machine translation (Meta M2M100); the English original follows.

Abstract

Diffusion models have emerged as the foundation of modern generative systems, yet their high memorization capacity raises privacy concerns. While differentially private (DP) training provides formal guarantees, it comes at a cost in sample quality. In this work, we study a complementary question: how much privacy loss is incurred during the sampling process itself? Directly analyzing modern neural denoisers is challenging because their sensitivity to a single training example is generally intractable. To overcome this challenge, we construct and analyze a clipped privacy-accountable empirical

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.27710, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange