Universal hidden monotonic trend estimation with contrastive learning

no code implementations • 18 Oct 2022 • Edouard Pineau, Sébastien Razakarivony, Mauricio Gonzalez, Anthony Schrapffer

In this paper, we describe a universal method for extracting the underlying monotonic trend factor from time series data.

Contrastive Learning Time Series +1

Time Series Source Separation with Slow Flows

no code implementations • 20 Jul 2020 • Edouard Pineau, Sébastien Razakarivony, Thomas Bonald

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.

blind source separation Time Series +1

Generalized mean shift with triangular kernel profile

no code implementations • 7 Jan 2020 • Sébastien Razakarivony, Axel Barrau

In order to situate this new class of methods in the general picture of the Mean Shift theory, we alo give a synthetic exposure of existing results of this field.

Clustering Visual Tracking