Exploiting the categorical reliability difference for binary classification

Published: 2018, Last Modified: 01 Apr 2026J. Frankl. Inst. 2018EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•A disparity index is proposed for measuring the reliability of matrices.•A study of the TER classifier and its regularized form in terms of matrix reliability.•A penalized TER classification algorithm is proposed.•Experimental evaluation based on 10 real-world data sets.
Loading