A New Method of Feature Selection for Flow Classification

Flow classification technology plays an important role in network router design, network security and network management etc. Network traffic data include a large number of relevant and redundant features, which will increased the flow classifier computational complexity, and affect the classification results. So the research on reducing the dimension of the network traffic data, and to find the important features of information-rich has important significance. In this paper, we provide an efficient approach for reduction the flow characteristics, namely Rough Set, and then construct traffic classifier in the feature subsets. The experimental results indicate that construction classifier on the reduction feature sets can not only obtain a higher computing performance, but also achieve a higher accuracy. (C) 2011 Published by Elsevier B.V. Selection and/or peer-review under responsibility of ICAPIE Organization Committee.