Thirdly and lastly, we also propose a method for estimating the performance of the classifier (scan detector) when labeled data is unavailable.If e+W alt; e, the true error rate in the stratum satisfies the user-specified requirements, correctional sampling can be stopped. If CM agt; fic\ then we have already corrected a sufficient number of instances, so the correctional sampling is complete.
|Title||:||Data Mining Techniques for Network Scan Detection|
|Author||:||Gyorgy J. Simon|
|Publisher||:||ProQuest - 2008|