Classification and Prediction Techniques Using Machine Learning for Anomaly Detection
Network traffic data can be classified into binary class (i.e. anomaly-free and all others) or multi-level classes (e.g., anomaly-free, likely to be anomaly-free, anomaly-free, anomaly, likely to be anomaly, and unable to determined). In this paper, the focus is on the common supervised learning algorithms and methods for binary classification. In the real world, it is possible that a data point belongs to more than one class or has similar attributes for multi-membership. In multi-level classification situation can be addressed by using multi-classification algorithms and then making decisions based on the membership functions acquired from the algorithms.