Work

Near Optimal Private and Robust Linear Regression

Year: 2023

Type: preprint

Abstract: We study the canonical statistical estimation problem of linear regression from $n$ i.i.d.~examples under $(\varepsilon,δ)$-differential privacy when some response variables are adversarially corrupted. We propose a variant of the popular differentially private stochastic gradient descent (DP-SGD) algorithm with two innovations: a full-batch gradient descent to improve sample complexity and a novel adaptive clipping to guarantee robustness. When there is no adversarial corruption, this algorithm improves upon the existing state-of-the-art approach and achieves a near optimal sample complexity. Under label-corruption, this is the first efficient linear regression algorithm to guarantee both $(\varepsilon,δ)$-DP and robustness. Synthetic experiments confirm the superiority of our approach.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.2301.13273

Authors: Liu, Xiyang, Jain, Prateek, Kong, Weihao, Oh, Sewoong, Suggala, Arun Sai

Institutions: University of Washington, Google (United States)

Language: English

Topic: Privacy-Preserving Technologies in Data

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: National Science Foundation

Awards: AI Institute for Future Edge Networks and Distributed Intelligence (AI-EDGE), Collaborative Research: MLWiNS: Physical Layer Communication revisited via Deep Learning, AI Institute: Institute for Foundations of Machine Learning, Scaling Laws of Deep Learning

OpenAlex ID: W4318902865

Cites:

Cited by: 3

Related to: 10