Abstract: Knowledge distillation has demonstrated considerable success in scenarios involving multi-class single-label learning. However, its direct application to multi-label learning proves ...