Generalized cross weighted Heronian mean operators and their application to binary classification

Abstract

Heronian mean (HM) operators, which can capture the interrelationship between input arguments with the same importance, have been a hot research topic as a useful aggregation technique. In this paper, we propose the generalized normalized cross weighted HM operators on the unit interval which can not only capture the interrelationships between input arguments but also aggregate them with different weights, some desirable properties are derived. Then, generalized cross weighted HM operators are extended to real number set and applied to binary classification. We list the detailed steps of binary classification with the developed aggregation operators, and give a comparison of the proposed method with the existing ones using the Iris dataset with 5-fold cross-validation (5-f cv), the accuracy of the proposed method for the training sets and the testing sets are both 100%.