English

Time Series Source Separation with Slow Flows

Machine Learning 2020-07-21 v1 Machine Learning

Abstract

In this paper, we show that slow feature analysis (SFA), a common time series decomposition method, naturally fits into the flow-based models (FBM) framework, a type of invertible neural latent variable models. Building upon recent advances on blind source separation, we show that such a fit makes the time series decomposition identifiable.

Keywords

Cite

@article{arxiv.2007.10182,
  title  = {Time Series Source Separation with Slow Flows},
  author = {Edouard Pineau and Sébastien Razakarivony and Thomas Bonald},
  journal= {arXiv preprint arXiv:2007.10182},
  year   = {2020}
}

Comments

INNF+ Workshop, ICML 2020

R2 v1 2026-06-23T17:14:59.744Z