Resampling Methods for Imbalanced Datasets in Multi-Label Classification: A Review

One of the most prevalent issues in the classification field is an imbalanced dataset. Because an instance can be assigned to more than one class label, multi-label categorization introduces complexity. The resampling technique is the most used approach to handle imbalanced problems because it is classifier-independent. The objective of this paper is to review resampling techniques that are already utilized to overcome the imbalanced issues in multi-label classification and to know what the insights and research gaps are. Some approaches are reviewed and analyzed to propose some challenges and research prospects.