Personalized Privacy Auditing and Optimization at Test Time

TL;DR

This paper investigates personalized privacy-preserving inference, demonstrating that individuals can often provide only a small subset of features without sacrificing model accuracy, thereby enhancing privacy and reducing verification efforts.

Contribution

It introduces a personalized, sequential feature selection algorithm that minimizes data sharing at test time while maintaining prediction accuracy.

Findings

Individuals can report as little as 10% of features without accuracy loss.

The proposed method reduces privacy risks and verification efforts.

Effective across multiple learning tasks.

Abstract

A number of learning models used in consequential domains, such as to assist in legal, banking, hiring, and healthcare decisions, make use of potentially sensitive users' information to carry out inference. Further, the complete set of features is typically required to perform inference. This not only poses severe privacy risks for the individuals using the learning systems, but also requires companies and organizations massive human efforts to verify the correctness of the released information. This paper asks whether it is necessary to require \emph{all} input features for a model to return accurate predictions at test time and shows that, under a personalized setting, each individual may need to release only a small subset of these features without impacting the final decisions. The paper also provides an efficient sequential algorithm that chooses which attributes should be provided by each individual. Evaluation over several learning tasks shows that individuals may be able to report as little as 10\% of their information to ensure the same level of accuracy of a model that uses the complete users' information.