Answers are grounded in this paper and cite their sources.
In Multi-label learning or Multi-label classification, each instance is associated with a set of classes. Current MLL studies show dataset characterization have an effect on the performance of certain MLL algorithm. Dataset understanding by exploratory data analysis is needed to identify MLL algorithms to be used. The study starts with the hypothesis that dataset linearity characteristic affects algorithm performance. The study identifies four commonly used MLL dataset with linear and non-linear characteristics. EPS with different base classifier was modified to test the hypothesis. Results shows that MLL linearly separated dataset are with better performance in EPS with linear base classifier. Also, EPS with non-linear base classifier works better in MLL dataset with non-linear in character.
No takes yet. Share an insight, caveat, or question.
Victoriano et al. (2019) studied this question.