ARTIFICIAL NEURAL NETWORKS - FOUNDATIONS AND APPLICATION TO A DECISION PROBLEM

With all the power of von Neumann serial computation, emulation of certain tasks performed routinely by humans has been painfully slow. Perhaps this is because there is evidence that human decision making results from massively parallel computation in the brain. This evidence has generated interest in constructing models of parallel computation which are called artificial neural networks. This paper outlines the fundamental architectures of artificial neural networks, details an artificial neural network which has potentially broad use, and reports on the results of its application to an important accounting problem that has been analyzed by other methods. The results indicate that the neural network model performed well compared to an inductive rule-generating algorithm and logistic regression.