An Attribute-Oriented Ensemble Classifier Based on Niche Gene Expression Programming

Ensemble of classifiers is a learning paradigm where a set of classifiers are jointly used to improve the classification accuracy. The main contributions of this paper include: (1) proposing a new concept named attribute selection set based on gene expression programming (GEP), (2) analyzing the principle of classifier ensemble, (3) proposing an attribute-oriented ensemble classifier Based on niche gene expression programming (AO-ECNG) to improve the accuracy of sub-classifiers and at the same time maintain the diversity among them, and (4) analyzing the relationship between predictive accuracy and ensemble size. Experimental results on 10 datasets suggest that AO-ECNG increases the predictive accuracy by 2.51%, 1.66%, 1.33% and 1 % respectively compared with single GEP-classifiers, Bagging, AdaBoost, and GEFS.