Seminar: Monika Mokrzycka, Institute of Plant Genetics, Polish Academy of Sciences
Approximation of the covariance matrix in doubly multivariate model
Seminar
Date:
Wednesday 26 February 2025Time:
13.00 – 14.00Location:
Campus Albano, Lecture room 26, hous 4, level 2Abstract
The aim of the talk is to present the methods of covariance matrix estimation by symmetric positive definite separable structure in the sens of the entropy loss function. Such estimation is of interest for data sets, where characteristics are measured repeatedly, e.g., in time or location. The approach can also be used as the method of measuring discrepancy between two distributions which differ in covariance matrix structures related to the study of power of the test. However, modern real-world experiments are producing more and more highdimensional data and then approximation via entropy loss function, which works under nonsingularity regime, is challenging.
Last updated: 2025-02-12
Source: Department of Statistics