A Unified View of Multi-Label Performance Measures
This work addresses the challenge of understanding algorithm performance across different measures in multi-label classification, which is incremental as it builds on existing measures.
The paper tackles the problem of evaluating multi-label classification algorithms by proposing a unified margin view that links label-wise and instance-wise margins to optimizing eleven performance measures, and introduces the LIMO max-margin approach with empirical verification.
Multi-label classification deals with the problem where each instance is associated with multiple class labels. Because evaluation in multi-label classification is more complicated than single-label setting, a number of performance measures have been proposed. It is noticed that an algorithm usually performs differently on different measures. Therefore, it is important to understand which algorithms perform well on which measure(s) and why. In this paper, we propose a unified margin view to revisit eleven performance measures in multi-label classification. In particular, we define label-wise margin and instance-wise margin, and prove that through maximizing these margins, different corresponding performance measures will be optimized. Based on the defined margins, a max-margin approach called LIMO is designed and empirical results verify our theoretical findings.