Performance Behavior of Intrusion Detection System (Ids) Based On Ensemble Base Classifier (EBC)
Rajesh Phursule

Dr. Rajesh Phursule, Associate Professor, Department of Computer Science, ICOER, Wagholi, Pune, India.
Manuscript received on 22 August 2019. | Revised Manuscript received on 07 September 2019. | Manuscript published on 30 September 2019. | PP: 3030-3034 | Volume-8 Issue-11, September 2019. | Retrieval Number: K21940981119/2019©BEIESP | DOI: 10.35940/ijitee.K2194.0981119
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: There is a tremendous growth in the area of information technology due to which, network defence is also facing major challenges. The conventional Intrusion Detection System (IDS) is not able to prevent the recent attacks and malwares. Hence, IDS which is an essential component of the network needs to be protected. Data mining introduce to the process of separate hidden, previously unknown and useful information from huge databases. Data Mining based Intrusion Detection System is combined with Multi-Agent System to improve the presentation of the IDS. We combine the classifiers which is the widespread approach, to increase the accuracy of a single classifier. For experimentation purpose, we use a benchmark intrusion detection dataset, which is KDDCup’99 and the accuracy of the classifiers were estimated using 10-fold cross validation method. In this work, we use the feature selection methods, namely Flexible mutual information based feature selection (FMIFS) and hybrid feature selection algorithm (HFS) to evaluate the importance of features. This work provides Support Vector Machine (SVM), Nave Bayes (NB) and Feed Forward Neural Network (FFNN) to classify attack and normal threads as well as to improve the accuracy we ensemble all classifier into single hybrid classifier using Bagging algorithm. The proposed hybrid approach achieves an accuracy rate of 95.11.
Keywords: IDS, KDDCup’99, FMIFS, HFS, SVM, FFNN
Scope of the Article: Classification