<?xml version="1.0" encoding="UTF-8"?><xml><records><record><source-app name="Biblio" version="7.x">Drupal-Biblio</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Luigi Grimaudo</style></author><author><style face="normal" font="default" size="100%">Marco Mellia</style></author><author><style face="normal" font="default" size="100%">Elena Baralis</style></author><author><style face="normal" font="default" size="100%">Ram Keralapura</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Self-Learning Classifier for Internet Traffic</style></title><secondary-title><style face="normal" font="default" size="100%">The 5th IEEE International Traffic Monitoring and Analysis Workshop (TMA 2013)</style></secondary-title></titles><dates><year><style  face="normal" font="default" size="100%">2013</style></year></dates><language><style face="normal" font="default" size="100%">eng</style></language><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Network visibility is a critical part of traffic engineering, network management, and security. Recently, unsupervised algorithms have been envisioned as a viable alternative&amp;nbsp;to automatically identify classes of traffic. However, the accuracy&amp;nbsp;achieved so far does not allow to use them for traffic classification&amp;nbsp;in practical scenario.&lt;br /&gt;In this paper, we propose SeLeCT, a Self-Learning Classifier&amp;nbsp;for Internet traffic. It uses unsupervised algorithms along with&amp;nbsp;an adaptive learning approach to automatically let classes of&amp;nbsp;traffic emerge, being identified and (easily) labeled. SeLeCT&amp;nbsp;automatically groups flows into pure (or homogeneous) clusters&amp;nbsp;using alternating simple clustering and filtering phases to remove&amp;nbsp;outliers. SeLeCT uses an adaptive learning approach to boost its&amp;nbsp;ability to spot new protocols and applications. Finally, SeLeCT&amp;nbsp;also simplifies label assignment (which is still based on some&amp;nbsp;manual intervention) so that proper class labels can be easily&amp;nbsp;discovered.&lt;br /&gt;We evaluate the performance of SeLeCT using traffic traces&amp;nbsp;collected in different years from various ISPs located in 3&amp;nbsp;different continents. Our experiments show that SeLeCT achieves&amp;nbsp;overall accuracy close to 98%. Unlike state-of-art classifiers, the&amp;nbsp;biggest advantage of SeLeCT is its ability to help discovering&amp;nbsp;new protocols and applications in an almost automated fashion.&lt;/p&gt;</style></abstract></record></records></xml>