The Importance of Being Discrete: Measuring the Impact of Discretization in End-to-End Differentially Private Synthetic Data

2025-04-09

AI Summary

This work addresses the imbalance between end-to-end differential privacy (DP) guarantees and utility in DP generative marginal models caused by discretization of continuous variables during preprocessing. We systematically evaluate and improve discretization strategies, quantifying—for the first time—the impact of uniform, quantile, k-means, and PrivTree discretization on utility and privacy risk in DP synthetic data generation. We propose three DP-enhanced discretizers (uniform, quantile, k-means) and an adaptive binning selection algorithm optimized for utility under DP constraints. Experiments demonstrate that our approach improves synthetic data utility by 29.7% on average, significantly mitigates membership inference attacks, reduces privacy budget consumption, and lowers computational overhead. The framework provides a provably private, efficient, and practical discretization paradigm for DP generative modeling.

Technology Category

• Machine Learning: Privacy
• Search and Optimization: Mixed Discrete/Continuous Search
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments

Abstract

Differentially Private (DP) generative marginal models are often used in the wild to release synthetic tabular datasets in lieu of sensitive data while providing formal privacy guarantees. These models approximate low-dimensional marginals or query workloads; crucially, they require the training data to be pre-discretized, i.e., continuous values need to first be partitioned into bins. However, as the range of values (or their domain) is often inferred directly from the training data, with the number of bins and bin edges typically defined arbitrarily, this approach can ultimately break end-to-end DP guarantees and may not always yield optimal utility. In this paper, we present an extensive measurement study of four discretization strategies in the context of DP marginal generative models. More precisely, we design DP versions of three discretizers (uniform, quantile, and k-means) and reimplement the PrivTree algorithm. We find that optimizing both the choice of discretizer and bin count can improve utility, on average, by almost 30% across six DP marginal models, compared to the default strategy and number of bins, with PrivTree being the best-performing discretizer in the majority of cases. We demonstrate that, while DP generative models with non-private discretization remain vulnerable to membership inference attacks, applying DP during discretization effectively mitigates this risk. Finally, we propose an optimized approach for automatically selecting the optimal number of bins, achieving high utility while reducing both privacy budget consumption and computational overhead.

Problem

• Evaluates discretization impact on DP synthetic data privacy
• Compares four DP discretization methods for optimal utility
• Proposes automated bin selection to enhance privacy and efficiency

Innovation

• DP versions of uniform, quantile, k-means discretizers
• PrivTree as best-performing DP discretizer
• Optimized automatic bin selection for high utility