In this paper, we presente a weighted loss function for multi-label classification with missing labels. The proposed approach show a clear improvement compared to original unweighted loss. The proposed approach is simple to integrate in pre-trained models, which is a relevant solution that has not been previously explored in the literature.
Karim Ibrahim, Elena Epure, Geoffroy Peeters, Gael Richard. Confidence-based Weighted Loss for Multi-label Classification with Missing Labels. The 2020 International Conference on Multimedia Retrieval (ICMR ‘20) , Jun 2020, Dublin, Ireland.