Unsupervised Deep One-Class Classification with Adaptive Threshold based on Training Dynamics
This work addresses a practical limitation in anomaly detection for scenarios where training data is contaminated, offering an incremental improvement over existing methods.
The paper tackles the problem of robust normality learning in deep one-class classification when abnormal samples are mixed into training data, proposing an unsupervised method that uses adaptive thresholding based on training dynamics to pseudo-label normal samples, resulting in improved anomaly detection performance on 10 benchmarks.
One-class classification has been a prevailing method in building deep anomaly detection models under the assumption that a dataset consisting of normal samples is available. In practice, however, abnormal samples are often mixed in a training dataset, and they detrimentally affect the training of deep models, which limits their applicability. For robust normality learning of deep practical models, we propose an unsupervised deep one-class classification that learns normality from pseudo-labeled normal samples, i.e., outlier detection in single cluster scenarios. To this end, we propose a pseudo-labeling method by an adaptive threshold selected by ranking-based training dynamics. The experiments on 10 anomaly detection benchmarks show that our method effectively improves performance on anomaly detection by sizable margins.