A Survey on the Choice Between Binary Classification and One-Class Classification  
Author Azadeh A. Zadeh


Co-Author(s) Joffrey L. Leevy; Taghi M. Khoshgoftaar


Abstract In a typical classification problem, the choice of machine learning strategy has a significant impact on performance results. When the target dataset has a class imbalance, the selection of the right strategy becomes even more important with respect to performance outcome. This survey paper focuses on the choice between a binary classification approach and a oneclass classification approach. Based on the results of previous studies, we determined that the preference for one approach over the other should be based not only on the degree of class imbalance and availability of minority class instances, but also on specific data complexity factors such as dataset modality. In addition, we discovered that none of these studies use big data for experimentation.


Keywords machine learning, one-class classification, binary classification, class imbalance
    Article #:  RQD27-74

Proceedings of 27th ISSAT International Conference on Reliability & Quality in Design
Virtual Event

August 4-6, 2022