LEARNING INVARIANT REPRESENTATIONS OF TIME-HOMOGENEOUS STOCHASTIC DYNAMICAL SYSTEMS - Archive ouverte HAL
Communication Dans Un Congrès Année : 2024

LEARNING INVARIANT REPRESENTATIONS OF TIME-HOMOGENEOUS STOCHASTIC DYNAMICAL SYSTEMS

Résumé

We consider the general class of time-homogeneous stochastic dynamical systems, both discrete and continuous, and study the problem of learning a representation of the state that faithfully captures its dynamics. This is instrumental to learning the transfer operator or the generator of the system, which in turn can be used for numerous tasks, such as forecasting and interpreting the system dynamics. We show that the search for a good representation can be cast as an optimization problem over neural networks. Our approach is supported by recent results in statistical learning theory, highlighting the role of approximation error and metric distortion in the learning problem. The objective function we propose is associated with projection operators from the representation space to the data space, overcomes metric distortion, and can be empirically estimated from data. In the discrete-time setting, we further derive a relaxed objective function that is differentiable and numerically well-conditioned. We compare our method against state-of-the-art approaches on different datasets, showing better performance across the board.
Fichier principal
Vignette du fichier
DPNets__post_submission_.pdf (4.15 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04555065 , version 1 (22-04-2024)

Identifiants

  • HAL Id : hal-04555065 , version 1

Citer

Vladimir R Kostic, Pietro Novelli, Riccardo Grazzi, Karim Lounici, Massimiliano Pontil. LEARNING INVARIANT REPRESENTATIONS OF TIME-HOMOGENEOUS STOCHASTIC DYNAMICAL SYSTEMS. ICLR 2024, May 2024, Vienna (AUSTRIA), Austria. ⟨hal-04555065⟩

Collections

X IP_PARIS
103 Consultations
189 Téléchargements

Partager

More