ABSTRACT
Deterministic neural networks such as backpropagation of error, multilayer perceptrons, and locally based radial basis methods have been a major focus of the neural network community in recent years. However, there has been a distinct, albeit less pronounced, interest in stochastic neural networks. In this review we provide the reader with a sense of the defining components of a stochastic neural network, as well as some of the issues arising from working with stochastic neural networks. In particular, issues revolving around hardware implementation, software simulation, and innovation are developed.
