Secure computing: logistic regression analysis with differential privacy

Yao Qiu; Shulan Wang; Caiguo Li; Guangdong Yang; Haiyan Wang; Fucai Luo; Yong Ding

doi:10.1117/12.2678974

12 May 2023 Secure computing: logistic regression analysis with differential privacy

Yao Qiu, Shulan Wang, Caiguo Li, Guangdong Yang, Haiyan Wang, Fucai Luo, Yong Ding

Proceedings Volume 12641, International Conference on Cryptography, Network Security, and Communication Technology (CNSCT 2023); 1264103 (2023) https://doi.org/10.1117/12.2678974
Event: International Conference on Cryptography, Network Security, and Communication Technology (CNSCT 2023), 2023, Changsha, China

Abstract

Logistic regression has the threat of privacy disclosure during data analysis. Differential privacy can prevent this security threat. However, differential privacy mechanism needs to add noise to the protected algorithm, which will affect the quality of data analysis. To improve the quality of data analysis, we propose dpLogic algorithm based on differential privacy and functional mechanism. This algorithm constructs an auxiliary function, which can help us get the noisy model as close as possible to the original model. Besides, dpLogic optimizes privacy budget allocation, reducing the disturbance of noise on the model. Experimental results illustrate that this algorithm can effectively improve the accuracy of data analysis at a high level of privacy protection.

Citation Download Citation

Yao Qiu, Shulan Wang, Caiguo Li, Guangdong Yang, Haiyan Wang, Fucai Luo, and Yong Ding "Secure computing: logistic regression analysis with differential privacy", Proc. SPIE 12641, International Conference on Cryptography, Network Security, and Communication Technology (CNSCT 2023), 1264103 (12 May 2023); https://doi.org/10.1117/12.2678974

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available