An insight into the standard back-propagation neural network model for regression analysis

This memorandum offers an extension of the work of Chiang et al., by investigating the issue of the standard backpropagation neural network model for regression analysis. It provides an analytical insight which shows how a regression result based on a simple back-propagation neural network may be adversely affected. It also proposes a general neural-network-based regression model, and demonstrates a practical example.