Apply nonlinear dimensionality reduction method to large-scale communication network traffic analysis
Weisong He, Zhiping Li, Hongmei Xiang · 2008
In this paper, we apply locally linear embedding method to large-scale communication network traffic analysis at flow-level. With the method, we can separate normal behavior from abnormal behavior by search k nearest neighbors. Compare with PCA, the method provide a nonlinear dimensionality reduction method for network traffic analysis and display more nonlinear information about data.