Title: Weak separability and L-separability for multi-way functional data
Authors: Kehui Chen - University of Pittsburgh (United States) [presenting]
Abstract: Multi-way functional data refers to functional data with double or multiple indices, such as brain-imaging data with spatial and temporal indices. To achieve efficient dimension reduction, one usually adopts the strong separability assumption that the covariance is a product of a spatial covariance and a temporal covariance. This assumption is quite restrictive and is often violated in real applications. We will introduce weak separability and L-separability, where the covariance can be approximated by a weighted sum of strong separable components. We will present the formal test procedure for the weak separability assumption and the algorithm for L-separable decomposition.