Privacy-Preserving Ridge Regression Without Garbled Circuits

Marc Joye

Paper 2017/732

Privacy-Preserving Ridge Regression Without Garbled Circuits

Marc Joye

Abstract

Ridge regression is an algorithm that takes as input a large number of data points and finds the best-fit linear curve through these points. It is a building block for many machine-learning operations. This report presents a system for privacy-preserving ridge regression. The system outputs the best-fit curve in the clear, but exposes no other information about the input data. This problem was elegantly addressed by Nikolaenko et al. (S\&P 2013). They suggest an approach that combines homomorphic encryption and Yao garbled circuits. The solution presented in this report only involves homomorphic encryption. This improves the performance as Yao circuits were the main bottleneck in the previous solution.

Metadata

Available format(s): PDF
Publication info: Preprint. MINOR revision.
Contact author(s): marc joye @ nxp com
History: 2017-08-01: received
Short URL: https://ia.cr/2017/732
License: CC BY

BibTeX

@misc{cryptoeprint:2017/732,
      author = {Marc Joye},
      title = {Privacy-Preserving Ridge Regression Without Garbled Circuits},
      howpublished = {Cryptology {ePrint} Archive, Paper 2017/732},
      year = {2017},
      url = {https://eprint.iacr.org/2017/732}
}