Relaxed exponential passivity criteria for memristor-based neural networks with leakage and time-varying delays

This paper investigates the problem of exponential passivity analysis for memristive neural networks with leakage and time-varying delays. Given that the input and output of the considered neural networks satisfy a prescribed passivity-inequality constraint, the more relaxed criteria are established in terms of linear matrix inequalities by employing nonsmooth analysis and Lyapunov method. The relaxations lie in three aspects: first, this obtained criteria do not really require all the symmetric matrices involved in the employed quadratic Lyapunov-Krasovskii functional to be positive definite; second, the activation functions become general; third, the time-varying delay is not needed to be differentiable. Finally, two numerical examples are given to show the effectiveness of the proposed criteria.