Incremental Input-to-State Stability criterion for adaptive Echo State Network systems

This paper presents a novel controller design strategy for an adaptive leaky-integrator Echo State Network (ESN) system based on the incremental input-to-state stability criterion. The online learning approach with the leaky-integrator echo state network system is developed through the recursive least square (RLS) algorithm. The controller design methods are derived and represented by a set of linear matrix inequality conditions. The effectiveness of the proposed controller design conditions is illustrated via a simulation example.