"Private Prediction Strikes Back!" Private Kernelized Nearest Neighbors with Individual R\'{e}nyi Filter

Yuqing Zhu, Xuandong Zhao, Chuan Guo, Yu-Xiang Wang

Keywords: private prediction, differential privacy, kNN

TL;DR: We propose a new private prediction mechanism that easily adapted to the changing private dataset and improve over existing private prediction methods.

Abstract: Most existing approaches of differentially private (DP) machine learning focus on private training. Despite its many advantages, private training lacks the flexibility in adapting to incremental changes to the training dataset such as deletion requests from exercising GDPR’s right to be forgotten. We revisit a long-forgotten alternative, known as private prediction, and propose a new algorithm named Individual Kernelized Nearest Neighbor (Ind-KNN). Ind-KNN is easily updatable over dataset changes and it allows precise control of the R\'{e}nyi DP at an individual user level --- a user's privacy loss is measured by the exact amount of her contribution to predictions; and a user is removed if her prescribed privacy budget runs out. Our results show that Ind-KNN consistently improves the accuracy over existing private prediction methods for a wide range of $\epsilon$ on four vision and language tasks. We also illustrate several cases under which Ind-KNN is preferable over private training with NoisySGD.