Aegis: A Correlation-Based Data Masking Advisor for Data Sharing Ecosystems
Omar Islam Laskar, Fatemeh Ramezani Khozestani, Ishika Nankani, Sohrab Namazi Nia, Senjuti Basu Roy, Kaustubh Beedkar
TL;DR
Aegis tackles the privacy-utility tradeoff in data-sharing ecosystems by avoiding exhaustive evaluation of masking configurations. It introduces a joint distribution reconstruction approach based on Iterative Proportional Fitting to estimate unobserved attribute–label distributions under masking, and a model-agnostic Predictive Utility Deviation metric computed from correlation measures such as Mutual Information, Chi-Square, and g3. The framework selects masking configurations that minimize utility loss, performing well with or without 1D data summaries, and shows an order-of-magnitude speedup over baselines while maintaining competitive downstream performance. Empirical results on three real-world datasets demonstrate robust reconstruction quality (TVD < 0.55 with summaries) and substantial efficiency gains, validating Aegis as a practical middleware for privacy-preserving data sharing. The work complements existing privacy techniques (e.g., k-anonymity, DP) by focusing on preserving predictive utility in a task-agnostic manner, enabling scalable, privacy-compliant data exchange for downstream ML tasks.
Abstract
Data sharing ecosystems connect providers, consumers, and intermediaries to facilitate the exchange and use of data for a wide range of downstream tasks. In sensitive domains such as healthcare, privacy is enforced as a hard constraint, any shared data must satisfy a minimum privacy threshold. However, among all masking configurations that meet this requirement, the utility of the masked data can vary significantly, posing a key challenge: how to efficiently select the optimal configuration that preserves maximum utility. This paper presents Aegis, a middleware framework that selects optimal masking configurations for machine learning datasets with features and class labels. Aegis incorporates a utility optimizer that minimizes predictive utility deviation, quantifying shifts in feature label correlations due to masking. Our framework leverages limited data summaries (such as 1D histograms) or none to estimate the feature label joint distribution, making it suitable for scenarios where raw data is inaccessible due to privacy restrictions. To achieve this, we propose a joint distribution estimator based on iterative proportional fitting, which allows supporting various feature label correlation quantification methods such as mutual information, chi square, or g3. Our experimental evaluation of real world datasets shows that Aegis identifies optimal masking configurations over an order of magnitude faster, while the resulting masked datasets achieve predictive performance on downstream ML tasks on par with baseline approaches and complements privacy anonymization data masking techniques.