Encrypted accelerated least squares regression
File(s)
Published version
Author(s)
Esperança, PM
Aslett, LJM
Holmes, CC
Type
Conference Paper
Abstract
Information that is stored in an encrypted format is, by definition, usually not amenable to statistical analysis or machine learning methods. In this paper we present detailed analysis of coordinate and accelerated gradient descent algorithms which are capable of fitting least squares and penalised ridge regression models, using data encrypted under a fully homomorphic encryption scheme. Gradient descent is shown to dominate in terms of encrypted computational speed, and theoretical results are proven to give parameter bounds which ensure correctness of decryption. The characteristics of encrypted computation are empirically shown to favour a non-standard acceleration technique. This demonstrates the possibility of approximating conventional statistical regression methods using encrypted data without compromising privacy.
Date Issued
2017
Date Acceptance
2017-01-01
Citation
Proceedings of Machine Learning Research, 2017, 54, pp.334-343
Publisher
PMLR
Start Page
334
End Page
343
Journal / Book Title
Proceedings of Machine Learning Research
Volume
54
Copyright Statement
© 2017 PLMR
Identifier
https://arxiv.org/abs/1703.00839
Source
Artificial Intelligence and Statistics (AISTATS)
Publication Status
Published
Start Date
2017-04-20
Finish Date
2017-04-22
Coverage Spatial
Fort Lauderdale, Florida, USA