Modelling multivariate spatio-temporal data with identifiable variational autoencoders

TL;DR

This paper extends identifiable variational autoencoders to nonlinear, nonstationary spatio-temporal blind source separation, demonstrating improved modeling and interpretation of complex multivariate data through simulations and meteorological applications.

Contribution

It introduces a novel nonlinear, identifiable variational autoencoder framework for spatio-temporal blind source separation, including methods for latent dimension estimation.

Findings

Effective nonlinear separation demonstrated in simulations

Improved meteorological data modeling and interpretation

Enhanced prediction accuracy using the proposed method

Abstract

Modelling multivariate spatio-temporal data with complex dependency structures is a challenging task but can be simplified by assuming that the original variables are generated from independent latent components. If these components are found, they can be modelled univariately. Blind source separation aims to recover the latent components by estimating the unmixing transformation based on the observed data only. The current methods for spatio-temporal blind source separation are restricted to linear unmixing, and nonlinear variants have not been implemented. In this paper, we extend identifiable variational autoencoder to the nonlinear nonstationary spatio-temporal blind source separation setting and demonstrate its performance using comprehensive simulation studies. Additionally, we introduce two alternative methods for the latent dimension estimation, which is a crucial task in order to obtain the correct latent representation. Finally, we illustrate the proposed methods using a meteorological application, where we estimate the latent dimension and the latent components, interpret the components, and show how nonstationarity can be accounted and prediction accuracy can be improved by using the proposed nonlinear blind source separation method as a preprocessing method.