Almost periodic attractor of delayed neural networks with variable coefficients

In this Letter, several sufficient conditions are obtained for the almost periodic attractor of delayed neural networks with variable coefficients based on the theory of exponential dichotomy, Lyapunov method and Halanay inequality technique. Many previous results reported in the literature are extended or improved for neural networks with or without delays. The obtained results are helpful to design globally stable almost periodic oscillatory neural networks. A numerical example is also given in the end to show the feasibility of our results.