期刊名称：International Journal of Computer and Information Technology
印刷版ISSN：2279-0764
出版年度：2013
卷号：2
期号：5
页码：942
出版社：International Journal of Computer and Information Technology
摘要：Ensemble learning is the machine learning paradigm concerned with utilizing multiple base classifiers which trained and then combined to achieve a strong generalization. This technique can be beneficial to semi- supervised learning, which exploits unlabeled data in addition to the labeled data, to achieve the best possible classification performance. One of the most common methods of semi-supervised learning is Self-training in which one base classifier is used to make decisions during a learning process. Instead of using just one base classifier into self- training process, an ensemble made up of three Support Vector Machines (SVM) classifiers with different kernels, which is denoted by SELF3SVM, is used in this paper. The experimental results with real and artificial data demonstrate that combining three SVM classifiers into self-training process is often much more accurate than the standard Self-training based on just one base classifier
