Novel stability criterion for DNNs via improved asymmetric LKF

Abstract

This paper briefly proposes an improved asymmetric Lyapunov-Krasovskii functional to analyze the stability issue of delayed neural networks (DNNs). By utilizing linear matrix inequalities (LMIs) incorporating integral inequality and reciprocally convex combination techniques, a new stability criterion is formulated. Compared to existing methods, the newly developed stability criterion demonstrates less conservatism and complexity in analyzing neural networks. To explicate the potency and preeminence of the proposed stability criterion, a renowned numerical instance is showcased, serving as an illustrative embodiment.

Keywords

• asymmetric lyapunov-krasovskii functional
• delayed neural network
• integral inequality
• stability criterion