Bayesian Perspective on Memorization and Reconstruction

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kaplan, Haim, Mansour, Yishay, Nissim, Kobbi, Stemmer, Uri
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918038303932416
author Kaplan, Haim
Mansour, Yishay
Nissim, Kobbi
Stemmer, Uri
author_facet Kaplan, Haim
Mansour, Yishay
Nissim, Kobbi
Stemmer, Uri
contents We introduce a new Bayesian perspective on the concept of data reconstruction, and leverage this viewpoint to propose a new security definition that, in certain settings, provably prevents reconstruction attacks. We use our paradigm to shed new light on one of the most notorious attacks in the privacy and memorization literature - fingerprinting code attacks (FPC). We argue that these attacks are really a form of membership inference attacks, rather than reconstruction attacks. Furthermore, we show that if the goal is solely to prevent reconstruction (but not membership inference), then in some cases the impossibility results derived from FPC no longer apply.
format Preprint
id arxiv_https___arxiv_org_abs_2505_23658
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Bayesian Perspective on Memorization and Reconstruction
Kaplan, Haim
Mansour, Yishay
Nissim, Kobbi
Stemmer, Uri
Cryptography and Security
Machine Learning
We introduce a new Bayesian perspective on the concept of data reconstruction, and leverage this viewpoint to propose a new security definition that, in certain settings, provably prevents reconstruction attacks. We use our paradigm to shed new light on one of the most notorious attacks in the privacy and memorization literature - fingerprinting code attacks (FPC). We argue that these attacks are really a form of membership inference attacks, rather than reconstruction attacks. Furthermore, we show that if the goal is solely to prevent reconstruction (but not membership inference), then in some cases the impossibility results derived from FPC no longer apply.
title Bayesian Perspective on Memorization and Reconstruction
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2505.23658