DATA QUALIFICATION - LOGIC ANALYSIS APPLIED TOWARD NEURAL-NETWORK TRAINING

For neural networks to develop good internal representations for patte rn mapping, noise in the training set data must be controlled. Because of the many difficulties associated with manually validating training data, we have focused on using decision table techniques as a practic al, domain-independent means of optimizing training set formulation. D ecision tables provide a variety of mechanisms whereby training set da ta can be processed to remove ambiguity, contradictions, and other noi se. In addition to serving as data filters, decision tables can be use d in the evaluation of neural network training.