Abstract
To handle various fuzziness in the input layer and the output layer of Distributed Single-layer Perceptron Network (DSPN), which is a valiant of conventional perceptron network, it is necessary to interpret noisy and/or subjective input data, and handle fuzzy teaching input and the respective fuzzy consequence, which have non-Boolean quantitative and/or qualitative meanings. This paper first proposes a fuzzy neural network and the learning method using fuzzy teaching input. As an application, Fuzzy Neural Expert System (FNES) for diagnosis of hepatobiliary disorders has been developed. We used a real medical database containing the results of nine biochemical tests of four hepatobiliary disorders. After learning by using training data (379 patients), the proposed system correctly diagnosed 77.3% of test (external) data from 163 previously unseen patients and correctly diagnosed 100% of the training data. Conversely, the diagnostic accuracy of the linear discriminant analysis was 63.2% of the test data and 67.0% of the training data. Neural network knowledge bases, which are generated by learning processes and training data accumulated in large scale medical databases, will be promising and useful knowledge acquisition tools in expert system development.
