Preprints, Working Papers, ... Year : 2024

Neural DDEs with Learnable Delays for Partially Observed Dynamical Systems

Abstract

Many successful methods to learn dynamical systems from data have recently been introduced. Such methods often rely on the availability of the system's full state. However, this underlying hypothesis is rather restrictive as it is typically not confirmed in practice, leaving us with partially observed systems. Utilizing the Mori-Zwanzig (MZ) formalism from statistical physics, we demonstrate that Constant Lag Neural Delay Differential Equations (ND-DEs) naturally serve as suitable models for partially observed states. In empirical evaluation, we show that such models outperform existing methods on both synthetic and experimental data.
Fichier principal
Vignette du fichier
main.pdf (3.27 Mo) Télécharger le fichier
LICENSE (11.09 Ko) Télécharger le fichier
README.md (81 B) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04715748 , version 1 (01-10-2024)

Identifiers

  • HAL Id : hal-04715748 , version 1

Cite

Thibault Monsel, Emmanuel Menier, Lionel Mathelin, Onofrio Semeraro, Guillaume Charpiat. Neural DDEs with Learnable Delays for Partially Observed Dynamical Systems. 2024. ⟨hal-04715748⟩
113 View
38 Download

Share

More