Using reservoir computing for dynamics forecast of noise-perturbed FitzHugh-Nagumo system

We investigate the capability of reservoir computing to predict the dynamics of excitable FitzHugh-Nagumo model, exposed to Gaussian white noise, and reproduce the phenomenon of coherence resonance in the reservoir. We train the neural network on the dynamics of the system with three noise amplitudes and then test in on different noises. We show that the reservoir computing can exhibit coherence resonance under external stimulus.