抄録
H-009
A Semi-Supervised MarginBoost Algorithm Applicable for Dissimilarity-Based Classifications
○Sang-Woon Kim・Thanh Binh Le（Myongji University）
Semi-supervised MarginBoost has been proposed for exploiting the unlabeled data in addition to the labeled data to improve performance on the classification task; performance improvement of any supervised learning algorithm with a multitude of unlabeled data. Designing DBCs using a semi-supervised boosting algorithm, by which the semi-supervised dissimilarity-based classification is implemented efficiently, is considered.