Modelling multivariate spatio-temporal data with identifiable variational autoencoders

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.

References in corpus (10)

• Auto-Encoding Variational Bayes
• Unsupervised Feature Extraction by Time-Contrastive Learning and Nonlinear ICA
• A dynamic nonstationary spatio-temporal model for short term prediction of precipitation
• Nonlinear ICA Using Auxiliary Variables and Generalized Contrastive Learning
• Spatial Blind Source Separation
• Disentangling Identifiable Features from Noisy Data with Structured Nonlinear ICA
• Blind source separation of tensor-valued time series
• Hidden Markov Nonlinear ICA: Unsupervised Learning from Nonstationary Time Series
• Nonlinear blind source separation exploiting spatial nonstationarity
• Vector Autoregressive Models with Spatially Structured Coefficients for Time Series on a Spatial Grid

Cited by in corpus (1)

• Identifiable Autoregressive Variational Autoencoders for Nonlinear and Nonstationary Spatio-Temporal Blind Source Separation