INTERPRETATION AND KNOWLEDGE DISCOVERY FROM A MULTILAYER PERCEPTRON NETWORK THAT PERFORMS WHOLE LIFE ASSURANCE RISK ASSESSMENT

This paper interprets the outputs from a Multilayer Perceptron (MLP) n etwork that performs a whole life assurance risk assessment task. Usin g a new method published by the first author, the paper finds the sign ificant, or key, inputs that the network uses to classify applicants f or whole life assurance into standard and non-standard risk. The ranki ng of the significant inputs enables the knowledge learned by the netw ork during training to be presented in the form of data relationships and induced rules which show that the network learns sensibly and effe ctively when compared with the training data set. This study demonstra tes the potential value of the knowledge discovery method for MLP netw ork validation and case-by-case interpretation both during network lea rning and network use. This has important implications for safety crit ical systems.