Adaptation of nonlinear systems through dynamic entropy estimation

Abstract

We show that adaptation and control of the target system can be achieved by linking the modification of its dynamical properties to the estimated difference of the distribution of the dynamical characteristics of the control and target signals. Subsequently the target system, which has initially different dynamical properties adjusts its dynamics via changes of its control parameters and synchronizes with the control one. The differences in the evolving probability distributions are evaluated through entropy estimation, causing the adaptation to be based solely on the statistical properties of the control and target signals without explicit knowledge of the underlying equations of the system.