Outlier detection in classification based on feature-selection-based regression

Abstract

The appearance of outliers results in a complexity to achieve an accurate classify. This paper aims to the detection and identification of outlier before selecting a suitable classifier. The problem is firstly converted to an high-dimensional regression, then we propose a novel method on combination of multiple-correlation-coefficientbased feature selection for dimensional reduction, t−test for sparsification, an iterated algorithm is also given. Performance on simulated numerical data and applications to low-dimensional iris data as well as high-dimensional DBWORD E-mail data demonstrate the superiority of the proposed method in outlier identification.