Machine Learning in Nonlinear Dynamical Systems

TL;DR

This paper reviews recent advances in applying machine learning to nonlinear dynamical systems, focusing on predicting future states and uncovering underlying dynamics from time-series data in an educational manner.

Contribution

It presents a comprehensive framework for using machine learning to analyze nonlinear systems, emphasizing both prediction and system identification.

Findings

ML frameworks can effectively predict nonlinear system evolution.

Unveiling analytical dynamics from data is feasible with proposed methods.

Educational approach makes complex concepts accessible.

Abstract

In this article, we discuss some of the recent developments in applying machine learning (ML) techniques to nonlinear dynamical systems. In particular, we demonstrate how to build a suitable ML framework for addressing two specific objectives of relevance: prediction of future evolution of a system and unveiling from given time-series data the analytical form of the underlying dynamics. This article is written in a pedagogical style appropriate for a course in nonlinear dynamics or machine learning.