Robust H ∞ Filter Design of Delayed Neural Networks

This paper is concerned with studying the robust H∞ filter design problem for a class of recurrent neural networks with timevarying delay. A delay-dependent criterion involving a scaling parameter is established under which the resulting filtering error system is globally asymptotically stable with a guaranteed performance in the H∞ sense. The purpose of the introduction of the scaling parameter lies in that the developed result can be efficiently applied to the neural networks with complex dynamic behaviors, which is illustrated by an example.