Robust Classification by a Nearest Mean-Median Rule for Generalized Gaussian pattern Distributions

摘要

To provide stability of classification,a robust supervised minimum distance salssifier based on the minimax(in the Huber sence)estimate of location isdesigned for the class of generalized Gaussian pattern dis variances,it is the nearest mean rule (NMean),and with relatively large variances,it is the nearest median rule (NMed).The proposed classifier exhibits good performance both under heavy-and short-Tailed pattern distributions.