Improving Differentially Private Models with Active Learning

Zhengli Zhao, Nicolas Papernot, Sameer Singh, Neoklis Polyzotis, Augustus Odena

Keywords: Differential Privacy, Active Learning

TL;DR: We introduce privacy-aware active learning techniques to improve differentially private classifiers and achieve state-of-the-art results.

Abstract: Broad adoption of machine learning techniques has increased privacy concerns for models trained on sensitive data such as medical records. Existing techniques for training differentially private (DP) models give rigorous privacy guarantees, but applying these techniques to neural networks can severely degrade model performance. This performance reduction is an obstacle to deploying private models in the real world. In this work, we improve the performance of DP models by fine-tuning them through active learning on public data. We introduce two new techniques - DiversePublic and NearPrivate - for doing this fine-tuning in a privacy-aware way. For the MNIST and SVHN datasets, these techniques improve state-of-the-art accuracy for DP models while retaining privacy guarantees.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:1910.01177/code)