Fig. 2: Deep-learning strategies for the forecast of noise-driven nonlinear instabilities.
From: Deep learning prediction of noise-driven nonlinear instabilities in fibre optics
Schematic of an artificial neural network (ANN) architecture for the analysis of incoherent modulation instability dynamics. The ANN is employed to predict the output spectrum and fluctuation properties of seeded modulation instability dynamics depending on the input seed parameters. Similarly, the properties of the fluctuating output spectrum (i.e. average spectrum and spectral correlation maps) can be used to infer the input seeds parameters. Both these ANN approaches can be compared to numerical simulations and experimental results considering incoherent modulation instability dynamics in nonlinear fibre propagation (see Methods).
