Session: Alumni Session

Date & Time: Saturday, May 23 — 1:30 PM–3:00 PM

Talk Title

Nested Testing of Covariance Separability in Two-Way Data via Empirical Likelihood

Abstract

Multivariate functional data with components defined over a common domain—such as regional BOLD signals in fMRI or multi-channel EEG time series—provide an important example of two-way data. A common simplifying assumption in their analysis is covariance separability, which enables more efficient computation and interpretation. Recent work has introduced two generalizations of this assumption, weak separability and partial separability, with the latter encompassing the former. In this talk, we propose a nonparametric, nested testing procedure based on empirical likelihood to assess the relative suitability of these separability structures for a random sample of two-way data. We establish theoretical properties of the proposed tests and illustrate their performance through applications to real data.