Combination of multiple diverse classifiers using belief functions for handling data with imperfect labels

• Published in: Expert systems with applications Vol. 39; no. 2; pp. 1698 - 1707
• Main Authors: Tabassian, Mahdi, Ghaderi, Reza, Ebrahimpour, Reza
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2012
• Subjects: Ambiguity; Belief functions framework; Classification; Classifier selection; Classifiers; Data with imperfect labels; Ensemble learning; Expert systems; Labels; Learning; Neural network; Neural networks; Training; Data with imperfect labels; Belief functions framework; Classifier selection; Neural network; Ensemble learning
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2011.06.061

► A new evidence-based ensemble model is proposed for handling data with imperfect labels. ► By utilizing the prototypes of the pre-defined classes, the possible uncertainty in the label of each learning data is detected. ► Based on the level of ambiguity concerning each learning sample class, it is...