PROVIDING UNDERSTANDING OF THE BEHAVIOR OF FEEDFORWARD NEURAL NETWORKS

The advent of artificial neural networks has stirred the imagination o f many in the field of knowledge acquisition. There is an expectation that neural networks will play an important role in automating knowled ge acquisition and encoding, however, the Problem solving knowledge of a neural network is represented at a subsymbolic level and hence is v ery difficult for a human user to comprehend. One way to provide an un derstanding of the behavior of neural networks is to extract their pro blem solving knowledge in terms of rules that can be provided to users , Several papers which propose extracting rules from feedforward neura l networks can be found in the literature, however, these approaches c an only deal with networks with binary inputs, Furthermore, certain ap proaches lack theoretical support and their usefulness and effectivene ss are debatable, Upon carefully analyzing these approaches, we propos e a method to extract fuzzy rules from networks with continuous-valued inputs, The method was tested using a real-life problem (decisionmaki ng by pilots involving combat situations) and found to be effective.