Received: 13 July 2018 Revised: 14 August 2018 Accepted: 26 August 2018
DOI: 10.1002/cpe.4999
SPECIAL ISSUE PAPER
Enhanced transductive support vector machine classification
with grey wolf optimizer cuckoo search optimization for
intrusion detection system
E.M. Roopa Devi
1
R.C. Suganthe
2
1
Department of Information Technology, Kongu
Engineering College, Erode, India
2
Department of Computer Science &
Engineering, Kongu Engineering College,
Erode, India
Correspondence
E.M. Roopa Devi, Assistant Professor
Department of Information Technology Kongu
Engineering College, Erode, India.
Email: roopasen5@gmail.com
Summary
These days, the Intrusion detection System (IDS) is the most talked topic among the scientist and
researchers and many research is going on in IDS, which is firmly connected to the protected
utilization of system administrations. IDS are an essential part of the security infrastructure.
The previous research works are focused to detect the attacks efficiently but it is failed to pro-
duce more accurate classification results. To stay away from the previously mentioned issues, in
the proposed framework, Hybrid Grey Wolf optimizer Cuckoo Search Optimization (HGWCSO)
along with Enhanced Transductive Support Vector Machine (ETSVM) is proposed. This exploration
incorporates the modules are, for example, preprocessing, selection of features and classification
of features. The processing of data is done by using normalization technique by using min-max
technique the main work is to replace the value missed and filters the features from NSL KDD
dataset values. The main objective of processing of data is to increase the accuracy of classifica-
tion. Then, the more relevant and optimal features are selected by using HGWCSO. The GWO
robustness and searching performance is increased by cuckoo search algorithm. Then, the classifi-
cation is performed to identify the intrusion attack types using ETSVM algorithm more efficiently.
This classification algorithm is used to improve the attack detection accuracy higher. The exper-
imental result concludes that the proposed HGWCSO with ETSVM algorithm provides better
performance metrics in terms of high precision, sensitivity, specificity, and accuracy than the
previous algorithms.
KEYWORDS
classification, data mining, ETSVM, feature selection, HGWCSO, intrusion detection
1 INTRODUCTION
Intrusion detection has constantly assumed a critical part in computer security research. Two general ways to deal with to intrusion detection are
such as misuse and anomaly detection. In misuse detection, essentially an example coordinating strategy, a client's exercises are contrasted and the
known mark example of meddlesome assaults. In anomaly detection, looks for patterns that depart from the normal attributes. Disregarding their
ability of identifying unknown attacks, anomaly detection system suffer from the essential trouble in characterizing what is normal properties.
1
Host based Intrusion Detection System (HIDS) is characterized as the way toward checking, gathering, and investigating occasions in a host to rec-
ognize interruptions, which are infringement of data frameworks security strategies, prompting ruptures in confidentiality, integrity or availability.
HIDS should thus be ready to determine such behavior.
Misuse detective work approach will faithfully establish intrusion attacks within the cause of well-known signatures of discovered vulnerabil-
ities. However, the protection department consultants raised interposition to outline rules or signature that limits the applying misuse detection
to create intelligent IDS. On the opposite hand, the bizarre person police investigation advance deals with the applied mathematics analysis and
pattern recognition issues. It is declared that this can be ready to notice novel attack with none previous knowledge for the classification model
which will extract invasion pattern and data throughout preparation.
2
Host based Intrusion Detection System (HIDS) is characterized as the way
Concurrency Computat Pract Exper. 2018;e4999. wileyonlinelibrary.com/journal/cpe © 2018 John Wiley & Sons, Ltd. 1 of 11
https://doi.org/10.1002/cpe.4999