Semi-supervised imbalanced multi-label classification with label propagation

Guodong Du, Jia Zhang, Ning Zhang, Hanrui Wu, Peiliang Wu, Shaozi Li

Abstract: Highlights•Provide a new method for semi-supervised imbalanced multi-label classification.•Propose a label regularization matrix to handle the imbalanced multi-label problem.•Leverage a collaborative manner to ensure the balanced outcomes.•Utilize weighted graph to exploit the representation of labeled and unlabeled data.•Experimental results demonstrate the superior performance of the proposed method.