Information gain feature selection for multi-label classification.

Abstract

In many important application domains, such as text categorization, biomolecular analysis, scene or video classification and medical diagnosis, instances are naturally associated with more than one class label, giving rise to multi-label classification problems. This fact has led, in recent years, to a substantial amount of research in multi-label classification. And, more specifically, many feature selection methods have been developed to allow the identification of relevant and informative features for multi-label classification. However, most methods proposed for this task rely on the transformation of the multi-label data set into a single-label one. In this work we have chosen one of the most wellknown measures for feature selection – Information Gain – and we have evaluated it along with common transformation techniques for the multi-label classification. We have also adapted the information gain feature selection technique to handle multi-label data directly. Our goal is to perform a thorough investigation of the performance of multi-label feature selection techniques using the information gain concept and report how it varies when coupled with different multi-label classifiers and data sets from different domains.