Forecasting of adaptive network’s dynamics

We investigate the possibility of forecasting the dynamics of the adaptive network, which topology changes in time, by using Reservoir Computing (RC). As an input signal, we use a signal averaged over 100 Kuramoto oscillators. Such macroscopic signal has a similar nature with EEG signal which is a macroscopic signal of a group of neurons that connections adapt in time. We find the optimal values of RC’s parameters for achieving maximal correlation between the output and the target signal.