Intrusion Detection System using Deep Neural Network and Regularization of Hyper Parameters with Adam Optimizer
R. Sekhar1, K. Thangavel2
1R. Sekhar, Dy Director, DRC, National Intelligence Grid, MHA, Bengaluru (Karnataka), India.
2K. Thangavel, Department of Computer Science, Periyar University, Salem (Tamil Nadu), India.
Manuscript received on 05 June 2019 | Revised Manuscript received on 14 June 2019 | Manuscript Published on 29 June 2019 | PP: 176-181 | Volume-8 Issue-5S, May 2019 | Retrieval Number: E10370585S19/19©BEIESP
Open Access | Editorial and Publishing 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: Intrusion Detection Systems (IDSs) study is unavoidable in the field of network security due to the present target oriented attacks for taking secret data of an organization. Classifying and detecting attacks are highly technical and tedious. In the existing models, the accuracy of intrusion detection in network traffic is different for different algorithms. This paper proposed a better intrusion detection system using Deep Neural Network with regularization of the hyper parameters. Adam optimization is proposed to optimize the weights in the neural network. The proposed system consists of six phases namely data collection, data framing, splitting of data for training and testing, pre-processing/encoding, regularization with Adam Optimizer, training and testing. It produces the better accuracy in detection process than the existing Deep Neural Network model. The bench mark data set NSL_KDD is collected and processed in the suggested system.
Keywords: Intrusion Detection Systems (IDSs), Deep Neural Network(DNN), Rectified Linear Unit (ReLU), Adaptive Moment Estimation (Adam) and Stochastic Gradient Decent (SGD).
Scope of the Article: Deep Learning