Anomaly detection identifies suspicious datasets in the rare observations as such datasets are significantly different from the standard datasets. The deep learning model acquires tremendous success in transforming many data mining tasks. The adaptability of deep learning models to deal with multivariate and high dimensional datasets helps in detecting anomalies. However, Deep learning models are difficult to detect intrusion instances when severely imbalanced datasets are available in big data.
   Deep learning with one class classification is the effective solution to tackle such issues. A one-class classification is an approach with a single class of known instances to identify the abnormal datasets. The one-class classification assists in extracting descriptive features to determine whether the data belongs to the known class or not. Exploiting deep feature extraction-based one-class classification produces optimal anomaly detection with high performance.