B1845
Title: Nonlinear blind source separation exploiting spatial nonstationarity
Authors: Mika Sipila - University of Jyvaskyla (Finland) [presenting]
Sara Taskinen - University of Jyvaskyla (Finland)
Klaus Nordhausen - University of Jyvaskyla (Finland)
Abstract: In spatial blind source separation the observed multivariate random fields are assumed to be mixtures of latent spatially dependent random fields. The objective is to recover latent random fields by estimating the unmixing transformation. Currently, the algorithms for spatial blind source separation can estimate only linear unmixing transformations. An identifiable variational autoencoder that can estimate nonlinear unmixing transformations is extended to spatially dependent data and its performance for both stationary and nonstationary spatial data is demonstrated using simulations.
