Instance-optimal Mean Estimation Under Differential Privacy
arXiv:2106.00463
Abstract
Mean estimation under differential privacy is a fundamental problem, but worst-case optimal mechanisms do not offer meaningful utility guarantees in practice when the global sensitivity is very large. Instead, various heuristics have been proposed to reduce the error on real-world data that do not resemble the worst-case instance. This paper takes a principled approach, yielding a mechanism that is instance-optimal in a strong sense. In addition to its theoretical optimality, the mechanism is also simple and practical, and adapts to a variety of data characteristics without the need of parameter tuning. It easily extends to the local and shuffle model as well.
References in corpus (9)
- Deep Learning with Differential Privacy
- Learning Differentially Private Recurrent Language Models
- Protection Against Reconstruction and Its Applications in Private Federated Learning
- Differentially Private Learning with Adaptive Clipping
- AdaCliP: Adaptive Clipping for Private SGD
- On the Power of Multiple Anonymous Messages
- A Primer on Private Statistics
- Locally Private Mean Estimation: Z-test and Tight Confidence Intervals
- New Bounds For Distributed Mean Estimation and Variance Reduction