In the domain of multi-label classification, label correlations play a crucial role in enhancing prediction precision. However, traditional methods heavily depend on ground-truth label sets, which can be incompletely tagged due to the diverse backgrounds of annotators and the significant cost associated with procuring extensive labeled datasets. To address these challenges, this paper introduces a novel multi-label classification method called updating Correlation-enhanced Feature Learning (uCeFL), which extracts label correlations directly from the data instances, circumventing the dependency on potentially incomplete label sets. uCeFL initially computes a revised label matrix by multiplying the incomplete label matrix with the label correlations extracted from the data matrix. This revised matrix is then utilized to enrich the original data features, enabling a neural network to learn correlation-enhanced representations that capture intricate relationships between data features, labels, and their interactions. Notably, label correlations are not static; they are dynamically updated during the neural network’s training process. Extensive experiments carried out on various datasets emphasize the effectiveness of the proposed approach. By leveraging label correlations within data instances, along with the hierarchical learning capabilities of neural networks, it offers a significant improvement in multi-label classification, even in scenarios with incomplete labels.
Label correlations take important roles in solving multi-label classification problems. However, many approaches extract label correlations from the corresponding label sets of data instances, which cannot provide satisfactory per-formance, because incompletely tagged label sets may exist in reality. To avoid such adverse effects, we directly extract label correlations from data instances and propose a multi-label approach called correlation-enhanced feature learning, named CeFL. First, a new revised label matrix is obtained by multiplying the incomplete label matrix with the label correlations, in which the label correlations are obtained from the data matrix rather than the incomplete label matrix. Then, the revised label information is used to expand the feature of the original data, and a neural network is used to learn the correlation-enhanced feature which contains relationships between data features, between labels and data features, and between labels. Compared with established approaches, ours is more intuitive and easier to understand, and label correlations obtained from the data can implicitly reflect the different importance of labels observed for identical data representation. Experiments on several databases demonstrate the effectiveness of our approach.