Timezone: »

 
Poster
Improving Sparse Vector Technique with Renyi Differential Privacy
Yuqing Zhu · Yu-Xiang Wang

Tue Dec 08 09:00 AM -- 11:00 AM (PST) @ Poster Session 1 #261
in Poster Session 2 »

The Sparse Vector Technique (SVT) is one of the most fundamental algorithmic tools in differential privacy (DP). It also plays a central role in the state-of-the-art algorithms for adaptive data analysis and model-agnostic private learning. In this paper, we revisit SVT from the lens of Renyi differential privacy, which results in new privacy bounds, new theoretical insight and new variants of SVT algorithms. A notable example is a Gaussian mechanism version of SVT, which provides better utility over the standard (Laplace-mechanism-based) version thanks to its more concentrated noise and tighter composition. Extensive empirical evaluation demonstrates the merits of Gaussian SVT over the Laplace SVT and other alternatives, which encouragingly suggests that using Gaussian SVT as a drop-in replacement could make SVT-based algorithms practical in downstream tasks.

Keywords: Audio and Speech Processing · Applications · Algorithms -> Sparsity and Compressed Sensing; Applications -> Computer Vision; Applications · Signal Processing; Optimization

Author Information

Yuqing Zhu (University of California Santa Barbara)
Yu-Xiang Wang (UC Santa Barbara)

More from the Same Authors