Continuous-Time Recurrent Neural Networks

Abstract—Thispaperstudiestheapproximationabilityofcon-tinuous-timerecurrentneuralnetworkstodynamicaltime-variantsystems.Itprovesthatanyfinitetimetrajectoryofagivendynam-icaltime-variantsystemcanbeapproximatedbytheinternalstateofacontinuous-timerecurrentneuralnetwork.Givenseveralspe-cialf