Semi-supervised multi-class Adaboost by exploiting unlabeled data

Research highlights We propose a semi-supervised learning method by using the multi-class boosting. It handles K-class classification without reducing into multiple two-class problems. The classifi...