OCFSP: Self-supervised One-class Classification Approach Using Feature-slide Prediction Subtask for Feature Data

Abstract One-class classification (OCC) is a machine learning problem where training data has only one class. Recently, self-supervised OCC algorithms have been increasing attention. These algorithms train the model for pretext tasks and use the model error for OCC. However, these tasks are specialized for images, and applying them to feature data is not practical or appropriate for such purpose. This paper proposes a one-class classification approach using feature-slide prediction (FSP) subtask for feature data (OCFSP). In particular, the self-labeled dataset is created from training data. In which additional feature vectors are generated by sliding original vectors and self-annotated as the number of the feature slide. Such a dataset is used to train a multi-class classifier, which aims to predict the number of the feature slides. Since this classification model is built using data from only one class, the FSP accuracy for seen data is high relative to unseen data. Accordingly, OCC could be made using the accuracy of FSP. Proposed methods are experimented with using the imbalanced-learn dataset.