ABSTRACT
Neural computing is a widely applied science. Since their revival in the mid 1980s artificial neural networks have been applied by researchers from a broad range of disciplines including traffic engineering. This chapter presents a case study that served to illustrate a number of important points and demonstrated that simulated data can be used effectively in preliminary development of a model. Quality and quantity of data are widely recognized as important issues in development of neural network models. In this case study the representation of different example types in the training set was shown to strongly influence performance. Development of a neural network approach to modelling traffic behaviour is difficult in the absence of a well structured and theoretically justified approach. The research detailed here has considered Learning Vector Quantization as an alternative. Numerous other paradigms exist and continue to be developed.
