DIRECT EXPLANATIONS FOR THE DEVELOPMENT AND USE OF A MULTI-LAYER PERCEPTRON NETWORK THAT CLASSIFIES LOW-BACK-PAIN PATIENTS
M.L. Vaughn, S. J. Cavill, Stewart J. Taylor, Michael A. Foy, Anthony J. B. Fogg · International Journal of Neural Systems · 2001
Using a new method published by the first author, this article shows how direct explanations can be provided to interpret the classification of any input case by a standard multilayer perceptron (MLP) network. The method is demonstrated for a real-world MLP that classifies low-back-pain patients into three diagnostic classes. The application of the method leads to the discovery of a number of mis-diagnosed training and test cases and to the development of a more optimal low-back-pain MLP network.