Mutual Optimization of Label Prediction and Node Representations for Multi-Label Node Classification
By bridging the representation and prediction spaces, MOLN effectively addresses the label-co-occurrence problem, achieving state-of-the-art performance on benchmark datasets with significant improvements.