The use of positive and diagonally dominant Lyapunov functions in analysis of recurrent neural networks

We demonstrate how to use a quadratic Lyapunov function V(x) = x'Px where P is positive definite and diagonally dominant to deduce the stability, l_2-l_2 induced gain and a reduced-order model for a class of discrete-time uncertain nonlinear systems which includes recurrent neural networks as a special case.

Y.C.Chu Cambridge University Engineering Dept.