Dynamic Time Lag Regression: Predicting What and When

Mandar Chandorkar (1) , Cyril Furtlehner (2) , Bala Poduval (3) , Enrico Camporeale (4, 5) , Michèle Sebag (6)

1 CWI - Centrum voor Wiskunde en Informatica
2 TAU - TAckling the Underspecified
3 UNH - University of New Hampshire
4 CIRES - Cooperative Institute for Research in Environmental Sciences
5 University of Colorado [Colorado Springs]
6 LRI - Laboratoire de Recherche en Informatique

This paper tackles a new regression problem, called Dynamic Time-Lag Regression (DTLR), where a cause signal drives an effect signal with an unknown time delay. The motivating application, pertaining to space weather modelling, aims to predict the near-Earth solar wind speed based on estimates of the Sun's coronal magnetic field. DTLR differs from mainstream regression and from sequence-to-sequence learning in two respects: firstly, no ground truth (e.g., pairs of associated sub-sequences) is available; secondly, the cause signal contains much information irrelevant to the effect signal (the solar magnetic field governs the solar wind propagation in the heliosphere, of which the Earth's magnetosphere is but a minuscule region). A Bayesian approach is presented to tackle the specifics of the DTLR problem, with theoretical justifications based on linear stability analysis. A proof of concept on synthetic problems is presented. Finally, the empirical results on the solar wind modelling task improve on the state of the art in solar wind forecasting.

Domaines: Apprentissage [cs.LG]; Réseau de neurones [cs.NE]

Mandar Chandorkar, Cyril Furtlehner, Bala Poduval, Enrico Camporeale, Michèle Sebag. Dynamic Time Lag Regression: Predicting What and When. ICLR 2020 - 8th International Conference on Learning Representations, Apr 2020, Addis Abeba, Ethiopia. ⟨hal-02422148⟩