Title: Effective semi-supervised approach towards intrusion detection system using machine learning techniques

Authors: Sharmila Kishor Wagh; Satish R. Kolhe

Addresses: Department of Computer Engineering, MES College of Engineering, Pune, Maharashtra, India ' School of Computer Science, North Maharashtra University, Jalgaon, Maharashtra, India

Abstract: Network security plays a very important role in today's web enabled world. In the 21st century, network traffic has increased because of the enormous growth in online users and their online communication. The number of security attacks is an increased with increase in internet users. The frequency and severity of such attacks have shown a great impact on network performance. Many classical machine learning algorithms have been proposed to solve the problem of intrusion detection with varying levels of success. Nowadays, availability of libelled data is a big issue. It is not only time consuming, but also expensive. Semi-supervised learning methods can make use of unlabelled examples in addition to the labelled ones. The developing field of semi-supervised learning, offers a promising direction for supplementary research. In this paper, we introduce a new semi-supervised mechanism for intrusion detection, which efficiently reduces false alarms, still maintaining a high detection rate. In our proposed semi-supervised learning approach, only a small quantity of labelled data and a large amount of unlabelled data has been used. This will improve the overall network security by reducing the security administrator's efforts and making the alert mechanism more practical.

Keywords: intrusion detection systems; semi-supervised learning; machine learning; entropy; network security; false alarms; detection rate.

DOI: 10.1504/IJESDF.2015.070395

International Journal of Electronic Security and Digital Forensics, 2015 Vol.7 No.3, pp.290 - 304

Received: 13 Sep 2014
Accepted: 23 Mar 2015

Published online: 04 Jul 2015 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article