Discovery of Single Independent Latent Variable
This addresses latent variable discovery for applications in data analysis, but it is incremental as it focuses on a special case of ICA where identifiability is achievable.
The paper tackles the problem of recovering a hidden independent component from an invertible mixture of two statistically independent sources, proposing an autoencoder with a discriminator that recovers the component up to entropy-preserving transformations, and demonstrates performance in tasks like image synthesis, voice cloning, and fetal ECG extraction.
Latent variable discovery is a central problem in data analysis with a broad range of applications in applied science. In this work, we consider data given as an invertible mixture of two statistically independent components and assume that one of the components is observed while the other is hidden. Our goal is to recover the hidden component. For this purpose, we propose an autoencoder equipped with a discriminator. Unlike the standard nonlinear ICA problem, which was shown to be non-identifiable, in the special case of ICA we consider here, we show that our approach can recover the component of interest up to entropy-preserving transformation. We demonstrate the performance of the proposed approach in several tasks, including image synthesis, voice cloning, and fetal ECG extraction.