Quantcast

Chaotifying linear Elman networks.

Research paper by Xiang X Li, Guanrong G Chen, Zengqiang Z Chen, Zhuzhi Z Yuan

Indexed on: 05 Feb '08Published on: 05 Feb '08Published in: IEEE transactions on neural networks / a publication of the IEEE Neural Networks Council



Abstract

A linear model of recurrent neural networks, called the Elman networks, is combined with the simple nonlinear modulo (mod) operation on its linear activated function so as to generate chaos purposely. Conditions on the weight matrix are obtained, under which the generated chaos satisfies the mathematical definition of chaos in the sense of T.Y. Li and J.A. Yorke (1975). Some simple and representative weight matrices are constructed for designing such Elman networks that can generate Li-Yorke chaos. Several numerical simulations are shown to verify and visualize the design.