Modelling multivariate spatio-temporal data with identifiable variational autoencoders
Sipilä, M., Cappello, C., De Iaco, S., Nordhausen, K., & Taskinen, S. (2025). Modelling multivariate spatio-temporal data with identifiable variational autoencoders. Neural Networks, 181, Article 106774. https://doi.org/10.1016/j.neunet.2024.106774
Published in
Neural NetworksDate
2025Copyright
© 2024 The Authors. Published by Elsevier Ltd.
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 unknown linear or nonlinear unmixing transformation based on the observed data only. In this paper, we extend recently introduced 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.
...
Publisher
ElsevierISSN Search the Publication Forum
0893-6080Keywords
Publication in research information system
https://converis.jyu.fi/converis/portal/detail/Publication/243505466
Metadata
Show full item recordCollections
Additional information about funding
We acknowledge the support from Vilho, Yrjö and Kalle Väisälä foundation for MS, the support from the Research Council of Finland (453691) to ST, the support from the Research Council of Finland (363261) to KN and the support from the HiTEc COST Action (CA21163) to KN and ST.License
Related items
Showing items with similar title or keywords.
-
A Bayesian spatio‐temporal analysis of markets during the Finnish 1860s famine
Pasanen, Tiia‐Maria; Voutilainen, Miikka; Helske, Jouni; Högmander, Harri (Wiley-Blackwell, 2022)We develop a Bayesian spatio-temporal model to study pre-industrial grain market integration during the Finnish famine of the 1860s. Our model takes into account several problematic features often present when analysing ... -
Spatio-temporal modeling of co-dynamics of smallpox, measles, and pertussis in pre-healthcare Finland
Pasanen, Tiia-Maria; Helske, Jouni; Högmander, Harri; Ketola, Tarmo (PeerJ Inc., 2024)Infections are known to interact as previous infections may have an effect on risk of succumbing to a new infection. The co-dynamics can be mediated by immunosuppression or modulation, shared environmental or climatic ... -
Spatio-temporal differences in the growth of wild and reared Atlantic salmon (Salmo salar L.) in the Baltic Sea
Peltola, Mikko (2013)Syönnösvaellus ja nopea kasvu ovat olennainen osa Atlantin lohen (Salmo salar L.) ekologiaa, niinpä kasvunopeuden selvittäminen on tärkeä osa kalatutkimusta. Tuottavimmat syönnösalueet tulisi tietää, jotta olisi mahdollista ... -
Of people and trees: exploring the spatio-temporal dynamics of urban and periurban dwellers’ social representations of trees.
Vuillot, Carole; Dufournet, Marylou; Prévot, Anne-Caroline (Open Science Centre, University of Jyväskylä, 2018)Almost 40 years ago, Pyle [1] started to warn the scientific community about the progressive disconnection between urban dwellers and nature. This so-called "extinction of experience" may affect individual relationships ... -
Statistical inference for eye movement sequences using spatial and spatio-temporal point processes
Ylitalo, Anna-Kaisa (University of Jyväskylä, 2017)Eye tracking is a widely used method for recording eye movements, which are important indicators of ongoing cognitive processes during the viewing of a target stimulus. Despite the variety of applications, the analyses ...