A critical overview of neural network pattern classifiers

摘要: A taxonomy of neural network pattern classifiers is presented which includes four major groupings. Global discriminant use sigmoid or polynomial computing elements that have 'high' nonzero outputs over most their input space. Local Gaussian other localized only a small region Nearest neighbor compute the distance to stored exemplar patterns and rule forming binary threshold-logic produce outputs. Results experiments are demonstrate provide error rates equivalent sometimes lower than those more conventional Gaussian. mixture, three using same amount training data. >