Global Robust Exponential Stability of Periodic Solution for Cohen-grossberg Type Neural Networks with Both Discrete and Neutral Time Varying Delays

A generalized Cohen-Grossberg neural network model with both discrete and neutral time varying delays is considered here. Activation functions and approximately behaved functions are not required to be differentiable. Global robust exponential stability of periodic solution is studied using a suitable Lyapunov functional by constructing a general eigen value problem [GEVP]. More generalized and less restrictive sufficient conditions are obtained in terms of linear matrix inequality (LMI).