作者: Huanhuan Mai , Xiaofeng Liao , Chuandong Li
DOI: 10.1016/J.CAM.2008.06.016
关键词: Constant (mathematics) 、 Mathematics 、 Artificial neural network 、 Applied mathematics 、 Numerical stability 、 Mathematical optimization 、 Exponential stability 、 Stability (learning theory) 、 Computation 、 Weighting 、 Variable (computer science)
摘要: In this paper, a new approach is proposed for stability issues of neutral-type neural networks (DNNs) with constant delay. First, the semi-free weighting matrices are and used instead known free to express relationship between terms in Leibniz-Newton formula simplify system synthesis obtain less computation demand. Second, global exponential conditions which conservative restrictive than results derived. At same time, based on above approach, fewer variable introduced construction Lyapunov functional augmented functional. Two examples given show their effectiveness advantages over others.