Logistic Regression Model for Differentially-Private Matrix Masking Data

Tóm tắt báo cáo

Abstract: In the age of data-driven innovation, privacy-protected data collection methods have become essential for balancing the interests of organizations and individuals. A recently proposed scheme utilizing local noise addition and matrix masking enables data collection while protecting individual privacy from all parties, including even the data manager. As such, statistical analysis on such privacy-preserved data is particularly challenging for nonlinear models. By leveraging a relationship between logistic regression and linear regression estimators, we propose the first valid method for estimating and making inferences from logistic regression under this setting. Theoretical analysis of the proposed estimators confirmed their validity; simulations and real data analyses demonstrate the superiority of the proposed estimators over naive logistic regression methods on privacy-preserved datasets.

Speaker: Dr. Linh Nghiem is a Lecturer in Statistics at the University of Sydney. His research focuses on developing novel methodologies for complex data settings, with interests in measurement error models, dimension reduction, and graphical models. He has published in leading journals such as Biometrika, JASA, Biometrics, and Statistica Sinica.