Chaos control on Universal Learning Network

A method is proposed for controlling chaotic phenomena on Universal Learning Network (U.L.N.). Generation and die-out of chaotic phenomena are realized by changing Lyapunov Number of U.L.N. to positive and negative value respectively, which is accomplished by adjusting U.L.N. parameters so as to minimize a criterion function that is the difference between the desired Lyapunov Number and its actual value. Both a gradient method utilizing second order derivatives of U.L.N. and a random search method are adopted to optimize the parameters.