|
Title:
|
FEATURE FILTERING TECHNIQUES APPLIED IN IP TRAFFIC CLASSIFICATION |
|
Author(s):
|
Michael Taynnan Barros, Reinaldo Cezar Gomes, Marcelo Sampaio de Alencar, Anderson Fabiano Costa |
|
ISBN:
|
978-989-8533-16-6 |
|
Editors:
|
Bebo White and Pedro Isaías |
|
Year:
|
2013 |
|
Edition:
|
Single |
|
Keywords:
|
Filtering, classification, IP Traffic, evaluation. |
|
Type:
|
Full Paper |
|
First Page:
|
227 |
|
Last Page:
|
234 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
IP traffic classifiers using machine learning algorithms are promising techniques for many applications involving management and network security. One of such applications is Traffic Engineering, with regard to the QoS problem, which is important to map an application into a service level. To develop a model for IP traffic classifiers using machine learning the set of features from a traffic sample needs to be defined properly. This paper presents an analysis of such problem invoking filtering techniques in order to decrease the training time and system complexity. The results show a reduction of the number of features to five, instead of common used 37, keeping a level of accuracy of more than 90%. |
|
|
|
|
|
|
