Anomaly detection using k-means clustering and decision tree classification

Mohd Zin, Hazlin (2016) Anomaly detection using k-means clustering and decision tree classification. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

	Text (24 Pages) Anomaly detection using k-means clustering and decision tree classification.pdf - Submitted Version Download (539kB)
	Text (Full text) Anomaly detection using k-means clustering and decision tree classification.pdf - Submitted Version Restricted to Registered users only Download (3MB)

Abstract

Nowadays, our country Malaysia was not exempt from cyber incidents. Although there are many types of security methods like access control, encryption, firewall are used but network security breaches increase day by day. With such unpredictable pattern of attacks, our defense calls for an urgent need to efficiently identify attacks and to classify them based on the degree of threats that they pose. One of the components of security that suit the ‘defense in depth’ model is called the Intrusion Detection System (IDS). IDS become an important defense to block for any network intrusion. An IDS is capable of detecting and sending early alarm upon risk exposure caused by any attack. A growing interest in the investigation of anomaly detection sparks from the ability of the approach to detect unknown attacks and to evaluate. A new hybrid mining approach is to improving current anomaly detection capabilities in IDS that would be securing an information infrastructure which is K-means clustering method and classification method. Thus, an urgent action needed to detect any attacks effectively. Data mining is the latest technology that been introduced in network security to fine regularities and irregularities in large data set. In this project, a hybrid data mining approach formed by combining the K-means clustering and classification. For the accuracy result, the detection and false alarm rate will be compared to the previous techniques that have been done before on the related research.

Actions (login required)

View Item