Bayesian Perspective on Memorization and Reconstruction

2025/05/29 by Kaplan, Haim, Mansour, Yishay, Nissim, Kobbi +1
#Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.2505.23658

Abstract

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.

Related

• Data Reconstruction: When You See It and When You Don't
• On the Privacy Effect of Data Enhancement via the Lens of Memorization
• No Prior, No Leakage: Revisiting Reconstruction Attacks in Trained Neural Networks
• Bayes' capacity as a measure for reconstruction attacks in federated learning
• Stolen Memories: Leveraging Model Memorization for Calibrated White-Box Membership Inference