Return to Colloquia & Seminar listing
PhD Exit Seminar: Variance Components Estimation under High-dimensional Linear Models
Special Events| Speaker: | Xiaohan Hu |
| Location: | 1147 MSB |
| Start time: | Thu, Jun 5 2025, 10:00AM |
Signal-to-noise ratio (SNR) estimation is a fundamental problem in high-dimensional statistics, with important applications in heritability analysis. This talk investigates the asymptotic properties of two widely used SNR estimators under various model assumptions. In the first part, we study the maximum likelihood (MLE) estimator derived from the Gaussian random effects model and establish its consistency and asymptotic normality under model misspecification, where the true coefficient vector is fixed and the noise components may be heterogeneous. In the second part, we extend the method-of-moments approach to multivariate linear models under both fixed and random effects settings. We derive the asymptotic distributions of the proposed estimators and develop inference procedures that accommodate residual heteroskedasticity.
Zoom link: https://ucdavis.zoom.us/j/92008370589?pwd=CiRw3KanAAOjqKzz9FQdMKqYnqTbBx.1 Meeting ID: 920 0837 0589 Passcode: 845531