Learning common and label-specific features for multi-Label classification with correlation information

•We extract common and label specific features simultaneously in learning process.•We adopt a novel assumption to exploit label correlations, that is, if two.labels are strongly correlated, their corresponding output tend to be similar.•In addition to label correlations, we also consider instance correlations by calculating instance similarity using K-Nearest Neighbor. In multi-label classification, many existing works only pay attention to the label-specific features and label correlation while they ignore the common features and instance correlation, which are also essential for building a competitive classifier. Besides, existing works usually depend on the assumption that they tend to have the similar label-specific features if two labels are correlated. However, this assumption cannot always hold in some cases. Therefore, in this paper, we propose a new approach of learning common and label-specific features for multi-label classification using the correlation information from labels and instances. First, we introduce l2,1-norm and l1-norm regularizers to learn common and label-specific features simultaneously. Second, we use a regularizer to constrain label correlations on label outputs instead of coefficient matrix. Finally, instance correlations are also considered through the k-nearest neighbor mechanism. Comprehensive experiments manifest the superiority of our proposed approach against other well-established multi-label learning algorithms for label-specific features.