Statistical Learning and Modelling for Complex Matrix-Valued Data
Conference
Proposal Description
This session will present recent statistical learning approaches for complex matrix-valued data, with emphasis on low-rank structure, dimension reduction, and the integration of informative covariates. The talks will cover methods for matrix completion and imputation of a partially observed response matrix
Y, factor-based reduction of matrix dimensions, and models that incorporate auxiliary covariate information, such as a matrix X. The session will also highlight emerging approaches, including neural network-based methods, for learning flexible relationships between matrix responses and covariates.