I. Introduction
Supervised and unsupervised learning are two fundamental learning paradigms. The main difference is in the manner in which data are utilized during learning. In supervised learning, the data set used comprises labeled data, whereas in unsupervised learning, only unlabeled data is used. Because, data collected from the real world is often unlabeled, the explosive growth of unlabeled data is a significant challenge to supervised learning. Hence, semisupervised learning has been the focus of considerable attention because its learning performance can be improved by incorporating information from both labeled and unlabeled data [1], [2].