Anomaly detection based on efficient Euclidean projection

摘要: Machine-learning algorithms are widely applied in traffic classification and anomaly detection. Due to the tremendous on network, an extremely challenging question arises: how efficiently accurately detect anomalous flow from backbone network. One solution is proposed, online anomaly-detection scheme, which based sparse feature selection method, Lasso. The can be solved by reformulating problem as optimization with i¾?1-ball constraint. At evaluation stage, authors preprocessed raw data trace trans-Pacific link between Japan United States generated set. Their empirical study shows that step quickly applying efficient Euclidean projection method; indeed, doing so resolves faster than using three classical i¾?1-min solvers. In terms of overall accuracy, true positive rate, false precision, F-measure, proposed scheme improves quality Copyright © 2015John Wiley & Sons, Ltd.