Benchmarking Multi-label Classification Algorithms

Multi-label classification is an approach to classification problems that allows each data point to be assigned to more than one class at the same time. Real life machine learning problems are often multi-label in nature—for example image labelling, topic identification in texts, and gene expression prediction. Many multi-label classification algorithms have been proposed in the literature and, although there have been some benchmarking experiments, many questions still remain about which approaches perform best for certain kinds of multi-label datasets. This paper presents a comprehensive benchmark experiment of eleven multilabel classification algorithms on eleven different datasets. Unlike many existing studies, we perform detailed parameter tuning for each algorithmdataset pair so as to allow a fair comparative analysis of the algorithms. Also, we report on a preliminary experiment which seeks to understand how the performance of different multi-label classification algorithms changes as the characteristics of multi-label datasets are adjusted.