Data Generation Using Geometrical Edge Probability for One-class Support Vector Machines

Abstract

In the matter of data that has only one class, a one-class support vector machine (OCSVM) model is one of the most widely used methods for anomaly detection as one class classification. However, the choice of hyperparameters, a critical step for learning effective OCSVM decision boundaries, remains open as a post analysis step and often undecided. To tackle this issue, this paper proposes a new data-generation method using geometrical edge probability suitable for OCSVM hyperparameter selection. It improves the limitations of existing methods in that the geometrical edge probability enables the generation of both pseudo target data and pseudo anomaly data. We evaluate the proposed method for datasets with 16 different dimensions and show the performance improvement.