Multi-label Classification: Inconsistency, Ambiguity and Class Balanced KNN Classification

Many existing researches employ one-vs-others approach to decompose a multi-label classification problem into a set of 2-class classification problems, one for each class. This approach is valid in traditional single-label classification. However, it incurs training inconsistency in multi-label classification, because a multi-label data point could belong to more than one class. In this work, we further develop classical K-Nearest Neighbor classifier and propose a novel Class Balanced K-Nearest Neighbor (BKNN) approach for multilabel classification by emphasizing balanced usage of data from all the classes. In addition, we also propose a Class Balanced Linear Discriminant Analysis approach to address high-dimensional multi-label input data. Promising experimental results on three broadly used multi-label data sets demonstrate the effectiveness of our approach.