Validation of Neural Networks Via Taxonomic Evaluation
Brian John Taylor, James T. Smith · Kluwer Academic Publishers eBooks · 2006
The purpose of the material in this chapter is to aid the V&V practitioner with the validation, of a proposed or implemented neural network-based system. Validation involves the assessment of the appropriateness of a particular neural network-based solution for the problem being solved This chapter has presented a taxonomic overview of neural network systems from multiple perspectives. Major characteristics of applications and their solution via neural networks have been presented. At the highest level, the issue addressed here is whether any neural network architecture could be an appropriate choice for the problem being solved. Given that a particular neural network model should be able to solve the problem, the next major concern is to focus upon the supporting requirements that should be addressed, e.g., the particular neural network architecture.