An edge prediction proposal for multi-label classification

Jose Luis Avila Jimenez, Sebastian Ventura

Keywords: Multi-label Classification, Edge prediction, Graph Neural Network

Abstract: Multi-label classification task, where each class can be assigned to several labels simultaneously, has been a growing research area during the last years, due to their ability to deal with many real worlds problems. Besides deep learning techniques have been extensively used in that context, where it ith worth highlighting Graph Neural Network as part of the deep learning specialised to cope with complex data. In the present work, a novedous multi-label classification technique is presented, based on using Graph Neural Network to learn the subjacent structure in multi-label datasets, and it is compared with others well-established multi-label methods.