ABSTRACT
In this chapter, the authors consider the task of anomaly detection under the presence of network big data. They review the existing methods of Support vector data description (SVDD) for anomaly detection under the presence of network big data. SVDD is considered as the main tool for big data anomaly detection due to its easy-to-understand geometrical interpretation. Due to this high effectiveness and the high accuracy of SVDD, it has been heavily investigated. SVDD has fruitful research outcomes that spread through four main research directions: theoretical improvement and discovery, improving the accuracy and efficiency of SVDD, and exploring its general application in different problems and areas. Besides the improvement of efficiency, SVDD is also capable of incorporating extra/latent information within the dataset to provide a more accurate and robust solution for anomaly detection. In network anomaly detection, heterogeneous datasets come from different aspects of the network. They are helpful in enhancing the accuracy and robustness of the final result.
