AN EVIDENTIAL EXTENSION OF THE MRII TRAINING ALGORITHM FOR DETECTING ERRONEOUS MADALINE RESPONSES

This paper integrates the evidential reasoning methodology with the pa rallel distributed learning paradigm of artificial neural networks (AN N), As such, this work presents an algorithm for the detection and, if possible, subsequent correction of the errors in the neuron responses in the output layer of the multiple adaptive linear element (MADALINE ) ANN, A geometrical perspective of the MADALINE ANN processing method ology is provided. This perspective is then used to formulate a statis tical specification to identify and quantify the sources of uncertaint ies in the MADALINE processing methodology, A new algorithm, EMRII, is then developed as an extension to the original MRII (MADELINE rule II ) algorithm, to formulate support and plausibility measures based on t he statistical specification, The support and plausibility measures, t hus formulated, are indicative of the degree of confidence of the ANN, in regards to the correctness of its outputs. Based on the support me asure, a scheme utilizing two thresholds is proposed to facilitate the interpretation of the support values for error prediction in the ANN responses, Finally, simulation results for the application of the EMRI I algorithm in the prediction of erroneous responses in an example pro blem is presented, These simulation results highlight the error detect ion capabilities of the EMRII algorithm.