An
Intelligent CRF Based Feature Selection for Effective Intrusion Detection
1Department of Information Science and Technology, Anna University, India
2Department of Computer Science and Engineering, University College of Engineering Tindivanam, India
Abstract: As the Internet applications are growing rapidly, the
intrusions to the networking system are also becoming high. In such a scenario,
it is necessary to provide security to the networks by means of effective
intrusion detection and prevention methods. This can be achieved mainly by
developing efficient intrusion detecting systems that use efficient algorithms
which can identify the abnormal activities in the network traffic and protect
the network resources from illegal penetrations by intruders. Though many
intrusion detection systems have been proposed in the past, the existing
network intrusion detections have limitations in terms of detection time and
accuracy. To overcome these drawbacks, we propose a new intrusion detection
system in this paper by developing a new intelligent Conditional Random Field
(CRF) based feature selection algorithm to optimize the number of features. In
addition, an existing layered approach based algorithm is used to perform
classification with these reduced features. This intrusion detection system
provides high accuracy and achieves efficiency in attack detection compared to
the existing approaches. The major advantages of this proposed system are
reduction in detection time, increase in classification accuracy and reduction
in false alarm rates.
Keywords: Intrusion detection
system, feature selection, false alarms, layered approach, intelligent CRF,
ICRFFSA, LAICRF.
Received January 31, 2013; accepted November 10, 2013