講演情報

[22p-A302-4]Model-free prediction of driven dynamical systems using reservoir computing

Kazuki Takashima1, 〇(P)Andre Roehm1, Takatomo Mihana1, Ryoichi Horisaki1, Makoto Naruse1 (1.Univ. of Tokyo)

キーワード:

reservoir computing、dynamical system、echo state network

We investigate an alternative way to predict the dynamics of driven systems. Instead of developing a mathematical model, we propose to use a simple reservoir computing system for model-free prediction. We demonstrate our scheme in a simple scenario, where we can successfully predict the response of a target system to novel stimuli.