Assimilation-based Learning of Chaotic Dynamical Systems from Noisy and Partial Data - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

Assimilation-based Learning of Chaotic Dynamical Systems from Noisy and Partial Data

Duong Nguyen
Said Ouala
Lucas Drumetz
Ronan Fablet

Résumé

Despite some promising results under ideal conditions (i.e. noise-free and complete observation), learning chaotic dy-namical systems from real life data is still a very challenging task. We propose a novel framework, which combines data assimilation schemes and neural network representation, namely Auto-Encoders and Ensemble Kalman Smoother, to learn the governing equations of dynamical systems. By treating the learning as a Bayesian estimation problem, our framework can deal with noisy and partial observations. Experiments on the chaotic Lorenz-63 dynamics with different noise settings demonstrate the advantages of our method over the state-of-the-art.
Fichier principal
Vignette du fichier
ICASSP2020(2).pdf (1.68 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02436060 , version 1 (12-01-2020)
hal-02436060 , version 2 (16-02-2020)

Identifiants

Citer

Duong Nguyen, Said Ouala, Lucas Drumetz, Ronan Fablet. Assimilation-based Learning of Chaotic Dynamical Systems from Noisy and Partial Data. ICASSP 2020 : International Conference on Acoustics, Speech, and Signal Processing, May 2020, Barcelona, Spain. ⟨10.1109/ICASSP40776.2020.9054718⟩. ⟨hal-02436060v2⟩
108 Consultations
224 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More