Dealing with a priori knowledge by fuzzy labels

Abstract The performances of two different estimators of a discriminant function of a statistical pattern recognizer are compared. One estimator is based on binary label values of the objects of the learning set (hard labels) and the other on continuous or multi-discrete label values in the interval [0,1] (fuzzy labels). By the latter estimator more detailed a priori knowledge of the contributing learning objects is used. In a discrete feature space, in which a multi-nomial distribution function has been assumed to exist, the expected classification error, based on fuzzy labels, can be more accurate than the one based on hard