Learning latent dynamics for partially observed chaotic systems
From MaRDI portal
Abstract: This paper addresses the data-driven identification of latent dynamical representations of partially-observed systems, i.e., dynamical systems for which some components are never observed, with an emphasis on forecasting applications, including long-term asymptotic patterns. Whereas state-of-the-art data-driven approaches rely on delay embeddings and linear decompositions of the underlying operators, we introduce a framework based on the data-driven identification of an augmented state-space model using a neural-network-based representation. For a given training dataset, it amounts to jointly learn an ODE (Ordinary Differential Equation) representation in the latent space and reconstructing latent states. Through numerical experiments, we demonstrate the relevance of the proposed framework w.r.t. state-of-the-art approaches in terms of short-term forecasting performance and long-term behaviour. We further discuss how the proposed framework relates to Koopman operator theory and Takens' embedding theorem.
Recommendations
- Linearly recurrent autoencoder networks for learning dynamics
- Koopman operator, geometry, and learning of dynamical systems
- Data-driven learning of nonautonomous systems
- Data-driven predictions of the Lorenz system
- Deep hidden physics models: deep learning of nonlinear partial differential equations
Cites work
- Data assimilation. Making sense of observations.
- Deterministic Nonperiodic Flow
- Discovering governing equations from data by sparse identification of nonlinear dynamical systems
- Embedology
- scientific article; zbMATH DE number 3810550 (Why is no real title available?)
- scientific article; zbMATH DE number 1145169 (Why is no real title available?)
- scientific article; zbMATH DE number 1953396 (Why is no real title available?)
- scientific article; zbMATH DE number 1866304 (Why is no real title available?)
- scientific article; zbMATH DE number 3050102 (Why is no real title available?)
- scientific article; zbMATH DE number 3109695 (Why is no real title available?)
- Identification of nonlinear systems using polynomial nonlinear state space models
- Least squares estimates in stochastic regression models with applications to identification and control of dynamic systems
- On dynamic mode decomposition: theory and applications
- On the identification of a vortex
- Turbulence, coherent structures, dynamical systems and symmetry.
Cited in
(20)- Learning to imitate stochastic time series in a compositional way by chaos
- A nonintrusive hybrid neural-physics modeling of incomplete dynamical systems: Lorenz equations
- Data-driven predictions of the Lorenz system
- Stochastic embeddings of dynamical phenomena through variational autoencoders
- LaSDI: parametric latent space dynamics identification
- Bounded nonlinear forecasts of partially observed geophysical systems with physics-constrained deep learning
- Data-driven forward and inverse problems for chaotic and hyperchaotic dynamic systems based on two machine learning architectures
- Using data assimilation to train a hybrid forecast system that combines machine-learning and knowledge-based components
- Data-driven learning of nonautonomous systems
- Bayesian framework for simulation of dynamical systems from multidimensional data using recurrent neural network
- Linearly recurrent autoencoder networks for learning dynamics
- A framework for machine learning of model error in dynamical systems
- Learning dynamical systems from data: a simple cross-validation perspective. IV: Case with partial observations
- A causality-based learning approach for discovering the underlying dynamics of complex systems from partial observations with stochastic parameterization
- Learning Theory for Dynamical Systems
- Geometry-preserving Lie group integrators for differential equations on the manifold of symmetric positive definite matrices
- Data-driven modeling of partially observed biological systems
- Reconstruction, forecasting, and stability of chaotic dynamics from partial data
- tLaSDI: thermodynamics-informed latent space dynamics identification
- Model reduction of dynamical systems with a novel data-driven approach: the RC-HAVOK algorithm
This page was built for publication: Learning latent dynamics for partially observed chaotic systems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5139802)
