Estimate of parameter-deviation degree for delayed neural networks

The stability of a neural network model may often be destroyed by the parameter deviations during the implementation. However, few results (if any) for the asymptotical stability of such system with a certain degree of parameter deviations have been reported in the literature. In this paper, we present a simple delayed neural network model, in which each parameter deviates the reference point with the degree no more than a certain value, and further investigate the robust asymptotical stability of this model and estimate the maximum permissible deviation degree.