Transfer learning for linear regression with differential privacy

Abstract Transfer learning, as a machine learning approach to enhance model generalization, has found widespread applications across various domains. However, the risk of privacy leakage during the transfer process remains a crucial consideration. Differential privacy, with its rigorous mathematical...

Full description

Saved in:
Bibliographic Details
Main Authors: Yiming Hou, Yunquan Song, Zhijian Wang
Format: Article
Language:English
Published: Springer 2024-12-01
Series:Complex & Intelligent Systems
Subjects:
Online Access:https://doi.org/10.1007/s40747-024-01759-8
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1832571204600332288
author Yiming Hou
Yunquan Song
Zhijian Wang
author_facet Yiming Hou
Yunquan Song
Zhijian Wang
author_sort Yiming Hou
collection DOAJ
description Abstract Transfer learning, as a machine learning approach to enhance model generalization, has found widespread applications across various domains. However, the risk of privacy leakage during the transfer process remains a crucial consideration. Differential privacy, with its rigorous mathematical foundations, has been proven to offer consistent and robust privacy protection. This study delves into the problem of linear regression transfer learning under differential privacy and, on this basis, proposes a novel strategy incorporating prior information as a constraint to further enhance model performance and stability. In scenarios where the transferable source is known, a two-step transfer learning algorithm incorporating prior information is proposed. This approach leverages prior knowledge to effectively constrain the model parameters, ensuring that the solution space remains reasonable throughout the transfer process. For cases where transferable sources are unknown, a non-algorithmic, cross-validation-based method for transferable source detection is introduced to mitigate adverse impacts stemming from non-informative sources. The effectiveness of the proposed algorithms is validated through simulations and real-world data experiments.
format Article
id doaj-art-e36222ad123a40bd87e5028ef0329023
institution Kabale University
issn 2199-4536
2198-6053
language English
publishDate 2024-12-01
publisher Springer
record_format Article
series Complex & Intelligent Systems
spelling doaj-art-e36222ad123a40bd87e5028ef03290232025-02-02T12:48:48ZengSpringerComplex & Intelligent Systems2199-45362198-60532024-12-0111111310.1007/s40747-024-01759-8Transfer learning for linear regression with differential privacyYiming Hou0Yunquan Song1Zhijian Wang2College of Science, China University of PetroleumCollege of Science, China University of PetroleumCollege of Science, China University of PetroleumAbstract Transfer learning, as a machine learning approach to enhance model generalization, has found widespread applications across various domains. However, the risk of privacy leakage during the transfer process remains a crucial consideration. Differential privacy, with its rigorous mathematical foundations, has been proven to offer consistent and robust privacy protection. This study delves into the problem of linear regression transfer learning under differential privacy and, on this basis, proposes a novel strategy incorporating prior information as a constraint to further enhance model performance and stability. In scenarios where the transferable source is known, a two-step transfer learning algorithm incorporating prior information is proposed. This approach leverages prior knowledge to effectively constrain the model parameters, ensuring that the solution space remains reasonable throughout the transfer process. For cases where transferable sources are unknown, a non-algorithmic, cross-validation-based method for transferable source detection is introduced to mitigate adverse impacts stemming from non-informative sources. The effectiveness of the proposed algorithms is validated through simulations and real-world data experiments.https://doi.org/10.1007/s40747-024-01759-8Linear regressionTransfer learningDifferential privacyLinear constraintsLasso
spellingShingle Yiming Hou
Yunquan Song
Zhijian Wang
Transfer learning for linear regression with differential privacy
Complex & Intelligent Systems
Linear regression
Transfer learning
Differential privacy
Linear constraints
Lasso
title Transfer learning for linear regression with differential privacy
title_full Transfer learning for linear regression with differential privacy
title_fullStr Transfer learning for linear regression with differential privacy
title_full_unstemmed Transfer learning for linear regression with differential privacy
title_short Transfer learning for linear regression with differential privacy
title_sort transfer learning for linear regression with differential privacy
topic Linear regression
Transfer learning
Differential privacy
Linear constraints
Lasso
url https://doi.org/10.1007/s40747-024-01759-8
work_keys_str_mv AT yiminghou transferlearningforlinearregressionwithdifferentialprivacy
AT yunquansong transferlearningforlinearregressionwithdifferentialprivacy
AT zhijianwang transferlearningforlinearregressionwithdifferentialprivacy