How more data can hurt: instability and regularization in next-generation reservoir computing

DOI10.1063/5.0262977MaRDI QIDQ6922752zbMATH OpenFDO

Authors Yuanzhao Zhang, Edmilson Roque dos Santos, Huixin Zhang, Sean P. Cornelius

Publication date 30 September 2025

Published in Chaos (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1063/5.0262977

Mathematics Subject Classification ID

Ordinary differential equations (34-XX) Dynamical systems and ergodic theory (37-XX)

Cites work

• A framework for machine learning of model error in dynamical systems
• A proposal on machine learning via dynamical systems
• Approximation bounds for random neural networks and reservoir systems
• Combining machine learning and data assimilation to forecast dynamical systems from noisy partial observations
• Controlling chaotic maps using next-generation reservoir computing
• Deep double descent: where bigger models and more data hurt*
• Discovering governing equations from data by sparse identification of nonlinear dynamical systems
• Do reservoir computers work best at the edge of chaos?
• Echo state networks are universal
• Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolation
• scientific article; zbMATH DE number 4074523 (Why is no real title available?)
• scientific article; zbMATH DE number 7164779 (Why is no real title available?)
• scientific article; zbMATH DE number 961607 (Why is no real title available?)
• Hybridizing traditional and next-generation reservoir computing to accurately and efficiently forecast dynamical systems
• Learning unseen coexisting attractors
• Model-free inference of unseen attractors: reconstructing phase space features from a single noisy trajectory using Reservoir computing
• Multifunctionality in a reservoir computer
• Nonlinear system identification. NARMAX methods in the time, frequency, and spatio-temporal domains
• Numerical Methods for Computing Angles Between Linear Subspaces
• On explaining the surprising success of reservoir computing forecaster of chaos? The universal machine learning dynamical system with contrast to VAR and DMD
• Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations
• Reconciling modern machine-learning practice and the classical bias-variance trade-off
• Recovering sparse networks: basis adaptation and stability under extensions
• Reservoir computing as digital twins for nonlinear dynamical systems
• The Generalization Error of Random Features Regression: Precise Asymptotics and the Double Descent Curve
• The size of the sync basin
• The size of the sync basin revisited
• Two models of double descent for weak features
• Using data assimilation to train a hybrid forecast system that combines machine-learning and knowledge-based components
• Using machine learning to predict statistical properties of non-stationary dynamical processes: System climate,regime transitions, and the effect of stochasticity
• Using machine learning to replicate chaotic attractors and calculate Lyapunov exponents from data

Cited in

(4)

• On the emergence of numerical instabilities in next generation reservoir computing
• A CFL-type condition and theoretical insights for discrete-time sparse full-order model inference
• Next-generation reservoir computing for dynamical inference
• Enhancing stability in next-generation reservoir computing by enforcing the echo-state property

This page was built for publication: How more data can hurt: instability and regularization in next-generation reservoir computing

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6922752)