Postprocessing for Iterative Differentially Private Algorithms

TL;DR

This paper introduces a post-processing method for iterative differential privacy algorithms that leverages intermediate outputs to enhance overall accuracy without compromising privacy guarantees.

Contribution

It presents a novel post-processing approach that uses intermediate outputs to improve the accuracy of differentially private iterative algorithms.

Findings

Improved accuracy in differentially private iterative algorithms

Effective utilization of intermediate outputs

Maintains privacy guarantees while enhancing results

Abstract

Iterative algorithms for differential privacy run for a fixed number of iterations, where each iteration learns some information from data and produces an intermediate output. However, the algorithm only releases the output of the last iteration, and from which the accuracy of algorithm is judged. In this paper, we propose a post-processing algorithm that seeks to improve the accuracy by incorporating the knowledge on the data contained in intermediate outputs.