Optimization in Control and Learning in Coupled Map Lattice Systems.

In this paper, we analyze various control algorithms that have been proposed for controlling spatiotemporal chaos in a globally coupled map lattice (CML) system. We reformulate the choice of feedback parameters in such systems as a constrained optimization problem and provide numerical and experimental results on the choice of optimal parameters for controlling the mean global Lyapunov exponent of a lattice. Finally, we propose a scheme to use this optimization technique to solve a learning problem in which such a CML system can be used to emulate the dynamics of an epileptic brain. [ABSTRACT FROM AUTHOR]