Chaos in a three dimensional neural network

An artificial neural network (ANN) consisting of three neurons has been con sidered. The equations of control are given by three differential equations (DE) with nonlinear, positive and bounded response functions of the neuron s. Bifurcation diagram and three dimensional (3-D) phase portraits of the m odel show rich dynamics. With the change in synaptic weight and decay rate, the system passes from stable to periodic and then chaotic regimes. Intere stingly, the system returns to periodic regime by further changing the syna ptic weight. Computer code to calculate the Lyapunov exponent (LE) has been written to confirm chaos. (C) 2000 Elsevier Science Inc. All rights reserv ed.