ゼミナール発表

日時: 7月17日(水)3限 (13:30-15:00)


会場: L1

司会: 加藤 有己
Sirikarn Pukkawannna 1161207: D, 中間発表 山口 英, 関 浩之, 安本 慶一, 門林 雄基
title: Unsupervised S-Transform based Network Anomaly Detection
abstract: Network anomalies, such as attacks, port scans, and equipment outages can have detrimental effects on Internet services. The most techniques for network anomalies detection requires prior knowledge and some of them cannot detect novel anomalies. For example, misuse detection techniques need pre-defined attack signatures to detect attacks and will not able to detect unknown (new) attacks. Supervised and semi-supervised detection techniques efficiently detect new attacks and anomalies, however, a labeled training data is required to be used for identifying anomalous events.
In this talk, a network anomaly detection method which can detect novel anomalies without ANY prior knowledge will be described. The method takes benefits of a signal processing technique called S-Transform to increase chances of discovering network anomalies which like needles buried in a haystack. Testing results on high-speed US-Japan traffic data and well-known evaluation traffic data for network intrusion detection systems will be shown in this talk.
language of the presentation: English