Machine Learning for Intrusion Detection
Lists
Only you see this
Abstract
Detection of unknown attacks in network traffic is gaining increasing importance as modern attacks are characterized by high variabilities and mutation rates. Traditional signature-based intrusion detection systems (IDS) are not able to detect unknown attacks due to failing availability of appropriate signatures. We present an alternative approach based on machine learning techniques which enable automatic construction of profiles for normal packet payloads and detection of deviations thereof. Experimental evaluation of our approach showed a remarkable detection accuracy at low false positive rates and a major improvement in comparison to the widely used open-source IDS Snort.
BibTeX
@inproceedings{laskov2007machine,
author = {Pavel Laskov and Konrad Rieck and Klaus-Robert Müller},
title = {{Machine Learning for Intrusion Detection}},
booktitle = {NATO ASI Mining Massive Data Sets for Security},
year = {2007},
url = {https://openreview.net/forum?id=1NiwGUQNdIU},
}Loading related content...