Reservoir Computing

Abstract

Reservoir Computing (RC) is an umbrella term for adaptive computational paradigms that rely on an excitable dynam- ical system, also called the ”reservoir.” The paradigms have been shown to be particularly promising for temporal signal processing. RC was also explored as a potential candidate for emerging nanoscale architectures. In this article we re- flect on the current state of RC and muse about its future. In particular, we propose a set of open problems that we think need to be addressed in order to make RC more main- stream.