TVineSynth: A Truncated C-Vine Copula Generator of Synthetic Tabular Data to Balance Privacy and Utility

2025-03-20

AI Summary

This paper addresses the challenge of balancing privacy preservation and data utility in synthetic tabular data generation. We propose TVineSynth, a privacy-enhanced generative framework based on truncated C-Vine copula structures. TVineSynth jointly optimizes vine tree truncation and directed bias injection to selectively eliminate privacy-leaking dependencies—such as those exploited by membership inference attacks (MIA) or attribute inference attacks (AIA)—while preserving statistical dependencies critical for downstream utility. Notably, it provides the first theoretical privacy guarantee against AIA under continuous sensitive attributes. Experiments on both synthetic and real-world datasets demonstrate that TVineSynth significantly outperforms state-of-the-art methods—with and without differential privacy—in resisting MIA and AIA, while improving average downstream prediction accuracy by 12.6%.

Technology Category

• Machine Learning: Privacy
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Bias, Fairness & Privacy

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

We propose TVineSynth, a vine copula based synthetic tabular data generator, which is designed to balance privacy and utility, using the vine tree structure and its truncation to do the trade-off. Contrary to synthetic data generators that achieve DP by globally adding noise, TVineSynth performs a controlled approximation of the estimated data generating distribution, so that it does not suffer from poor utility of the resulting synthetic data for downstream prediction tasks. TVineSynth introduces a targeted bias into the vine copula model that, combined with the specific tree structure of the vine, causes the model to zero out privacy-leaking dependencies while relying on those that are beneficial for utility. Privacy is here measured with membership (MIA) and attribute inference attacks (AIA). Further, we theoretically justify how the construction of TVineSynth ensures AIA privacy under a natural privacy measure for continuous sensitive attributes. When compared to competitor models, with and without DP, on simulated and on real-world data, TVineSynth achieves a superior privacy-utility balance.

Problem

• Balances privacy and utility in synthetic data generation
• Uses vine copula model to control data approximation
• Ensures privacy against membership and attribute inference attacks

Innovation

• Uses vine copula for synthetic data generation
• Balances privacy and utility via truncation
• Targets bias to zero out privacy-leaking dependencies