Analysis of Complex Valued Recurrent Neural Network with Time-Delays

Ritu Vishwakarma

Abstract: We have presented a complex valued recurrent neural network given by a set of non-linear differential equation. In this paper, activation dynamics of complex valued neural networks are studied on general time scales. Besides presenting conditions guaranteeing the existence of a unique equilibrium pattern, its global exponential stability is discussed. Some numerical examples for different time scales are given in order to highlight the results. The equation is shown to possess a unique equilibrium point beside having bounded solution. The equation is shown to possess a unique equilibrium point beside having a bounded solution.

Keywords: Complex valued recurrent neural network, unique equilibrium point, contraction mapping, stability, activation function