Learn to synchronize, synchronize to learn

In recent years, the artificial intelligence community has seen a continuous interest in research aimed at investigating dynamical aspects of both training procedures and machine learning models. Of particular among recurrent neural networks, we have Reservoir Computing (RC) paradigm characterized by conceptual simplicity fast scheme. Yet, guiding principles under which RC operates are only partially understood. this work, analyze role played Generalized Synchronization (GS) when to solve generic task. particular, show how GS allows reservoir correctly encode system generating input signal into its dynamics. We also discuss necessary sufficient conditions for be feasible approach. Moreover, explore that ergodicity plays process, showing presence outcome apply multiple trajectories. Finally, satisfaction can measured means mutual false nearest neighbors index, makes effective practitioners theoretical derivations.