Multi-label learning aims at predicting potentially multiple labels for a given instance. Conventional multi-label learning approaches focus on exploiting the label correlations to improve the accurac
