Intrusion Detection System for Large Scale Data using Machine Learning Algorithms
Sayali R. Kshirsagar1, P.B.Kumbharkar2

1Sayali R. Kshirsagar, M.E. Computer Engineering from JSPM’s Rajarshi Shahu College of Engineering Pune (Maharashtra), India.
2P. B. Kumbharkar, Professor in Computer Engineering, Dean (Planning and Development) and IQAC CO-ordinator, Rajarshi Shahu College of Engineering Tathawade Pune (Maharashtra), India.
Manuscript received on July 20, 2019. | Revised Manuscript received on August 10, 2019. | Manuscript published on August 30, 2019. | PP: 706-711 | Volume-8 Issue-6, August 2019. | Retrieval Number: F7971088619/2019©BEIESP | DOI: 10.35940/ijeat.F7971.088619
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: To provide security to internet assets, Intrusion Detection System (IDS) is most essential constituent. Due to various network attacks it is very hard to detect malicious activities from remote user as well as remote machines. In such a manner it is mandatory to analyze such activities which are normal or malicious. Due to insufficient background knowledge of system it is hard to detect malicious activities of system. In this work we proposed intrusion detection system using various soft computing algorithms, the system has categorized into three different sections, in first section we execute the data preprocessing as well as generate background knowledge of system according to two training data set as well as combination genetic algorithm. Once the background knowledge has generated system executes for prevention mode. In prevention mode basically it works for defense mechanism from various networks and host attacks. System uses two data sets which contain around 42 attributes. The system is able to support for NIDS as well as HIDS respectively. The result section will show how proposed system is better than classical machine learning algorithms. With the help of various comparative graphs as well as detection rate of systems we conclude proposed system provides the drastic supervision in vulnerable network environment. The average accuracy of proposed system is 100% for DOS attacks as well as around more than 90% plus accuracy for other as well as unknown attacks respectively.
Keywords: Genetic Algorithm, HIDS Machine Learning Algorithm, NIDS, Ensemble method.