ABSTRACT
This chapter discusses the recurrent neurons and networks that may had questions about the stability of certain complex dynamical systems. In fact, the main criterion for such a network to be useful is its stability under noisy inputs and external perturbations. The idea of recurrent neurons and recurrent neural networks (NNs) has been extended to model more complex phenomena such as chaotic systems. The concept of radial basis functions (RBFs) is perhaps the most significant advancement made in the field of NN structures after Rosenblatt introduced the concept of a perceptron, where a collection of activation functions was organized to form a layered network to approximate a nonlinear function. An NN can cover a continuum from full knowledge to no knowledge of the structure of the process to be identified. However, it is very common to see standard NN structures being applied no matter to what degree the structure of the dynamics is known.
