A Pairwise Homogeneity Active Learning

Machine Learning has becoming an emerging topic within data mining. The active learning is also an upcoming topic. Active learning is the process of labeling unlabeled data instances by using queries and the labeling process will be done by expert labelers such as an oracle. The process of labeling will be very expensive and time consuming. The proposed method called a pairwise homogeneity active learning method which is an unsupervised label refinement method by using a pairwise homogeneity between the pair of data instances which improves the quality of the label. In this method we use a non expert labeler to provide the class label for data instances. The experimental results shows that the proposed method improves the labeling quality of the data instances which are being labeled.

Keywords: active learning, data mining, labeling, pairwise homogeneity, data instances.