Active anomaly detection based on deep one-class classification

Active learning has been utilized as an efficient tool in building anomaly detection models by leveraging expert feedback. In active framework, a model queries samples to be labeled experts and re-trains the with data samples. It unburdens obtaining annotated datasets while improving performance. However, most of existing studies focus on helping identify many abnormal possible, which is sub-optimal approach for one-class classification-based deep detection. this paper, we tackle two essential problems Deep SVDD: query strategy semi-supervised method. First, rather than solely identifying anomalies, our selects uncertain according adaptive boundary. Second, apply noise contrastive estimation training classification incorporate both normal effectively. We analyze that proposed loss individually improve process further when combined together seven datasets.