Dealing with Imbalanced Dataset Leveraging Boundary Samples Discovered by Support Vector Data Description

These days, imbalanced datasets, denoted throughout the paper by ID, (a dataset that contains some (usually two) classes where one considerably smaller number of samples than other(s)) emerge in many real world problems (like health care systems or disease diagnosis systems, anomaly detection, fraud stream based malware detection and so on) these datasets cause under-training minority class(es) over-training majority class(es), bias towards classification process application. Therefore, take focus researchers any science there are several solutions for dealing with this problem. The main aim study IDs is to resample borderline discovered Support Vector Data Description (SVDD). There naturally two kinds resampling: Under-sampling (U-S) over-sampling (O-S). O-S may occurrence over-fitting (the its drawback). U-S can significant information loss In study, avoid drawbacks sampling techniques, we on be misclassified. data points misclassified considered which border(s) between class(es). First SVDD, find examples; then, resampling applied over them. At next step, base classifier trained newly created dataset. Finally, compare result our method terms Area Under Curve (AUC) F-measure G-mean other state-of-the-art methods. We show has better results methods experimental study.