Self-Train LogitBoost for Semi-supervised Learning

• Published in: Engineering Applications of Neural Networks Vol. 517; pp. 139 - 148
• Main Authors: Karlos, Stamatis, Fazakis, Nikos, Kotsiantis, Sotiris, Sgarbas, Kyriakos
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2015; Springer International Publishing
• Series: Communications in Computer and Information Science
• Subjects: Classification method; Computing: general; Data mining; Labeled and/or unlabeled data; Logitboost; Pattern recognition; Semi-supervised learning
• ISBN: 9783319239811; 3319239813
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-3-319-23983-5_14

Semi-supervised classification methods are based on the use of unlabeled data in combination with a smaller set of labeled examples, in order to increase the classification rate compared with the supervised methods, in which the total training is executed only by the usage of labeled data. In this work, a self-train Logitboost algorithm is presented. The self-train process improves the results by using the accurate class probabilities for which the Logitboost regression tree model is more confident at the unlabeled instances. We performed a comparison with other well-known semi-supervised classification methods on standard benchmark datasets and the presented technique had better accuracy in most cases.