Dimensionality reduction using stacked Kernel Discriminant Analysis for multi-label classification

Multi-label classification in which each instance may belong to more than one class is a challenging research problem. Recently, a considerable amount of research has been concerned with the development of "good" multi-label learning methods. Despite the extensive research effort, many scientific challenges posed by e.g. curse-of-dimensionality and correlation among labels remain to be addressed. In this paper, we propose a new approach to multi-label classification which combines stacked Kernel Discriminant Analysis using Spectral Regression (SR-KDA) with state-of-the-art instance-based multi-label (ML) learning method. The proposed system is validated on two multi-label databases. The results indicate significant performance gains when compared with the state-of-the art multi-label methods for multi-label classification.