Exponential stability of neural networks with asymmetric connection weights (Q2468109)

The paper investigates the exponential stability of a class of neural networks with asymmetric connection weights of the form \[ \frac{du}{dt}=-u+P_{\Omega }(Wu+\alpha q), \] where \(\alpha >0,u,q\in \mathbb{R}^{\ell }, W=E-\alpha M,E=\)identity matrix, \(M\) is an \(\ell \times \ell \) matrix, \(\Omega =\{u\in \mathbb{R}^{\ell };d_{i}\leq u_{i}\leq h_{i},i\in I\},\;I\subset L:=\{1,2,...,\ell \}\) and \(P_{\Omega }:\mathbb{R}^{\ell }\rightarrow \Omega \) is a projection operator, defined by \(P_{\Omega }\left( u\right) =\left[ P_{\Omega }(u_{1}),...,P_{\Omega }\left( u_{\ell }\right) \right] ^{T},\) with \(P_{\Omega }\left( u_{i}\right) =u_{i}\) for \( i\in L-I\), and \(P_{\Omega }(u_{i})=\min \left( h_{i},\max \left( u_{i},d_{i}\right) \right) \) for \(i\in I\) .