Dynamical analysis of a parameter-aware reservoir computer

Abstract

Reservoir computing has been shown to be a useful framework for predicting critical transitions of a dynamical system if the bifurcation parameter is also provided as an input. Its utility is significant because in real-world scenarios, the exact model equations are unknown. This Letter shows how the theory of dynamical system provides the underlying mechanism behind the prediction. Using numerical methods, by considering dynamical systems which show Hopf bifurcation, we demonstrate that the map produced by the reservoir after a successful training undergoes a Neimark-Sacker bifurcation such that the critical point of the map is in immediate proximity to that of the original dynamical system. In addition, we have compared and analyzed different structures in the phase space. Our findings provide insight into the functioning of machine learning algorithms for predicting critical transitions.

References in corpus (10)

• Emergence of scaling in random networks
• Recent Advances in Physical Reservoir Computing: A Review
• Next Generation Reservoir Computing
• Attractor Reconstruction by Machine Learning
• Amplitude Death: The emergence of stationarity in coupled nonlinear systems
• Solving Fokker-Planck equation using deep learning
• Using reservoir computers to distinguish chaotic signals
• Experimental observation of a transition from amplitude to oscillation death in coupled oscillators
• Finite-time Lyapunov exponents of deep neural networks
• Criticality in Reservoir Computer of Coupled Phase Oscillators

Cited by in corpus (2)

• Finite-time Lyaponov analysis of a trained reservoir computer
• Dynamics of reservoir computing for crises prediction