Intrusion Detection System using Binary and Multiclass Deep Neural Network Classification
Mohammed Ishaque, Md Gapar Md Johar, Ali Khatibi, Mohammad Yamin
Abstract
Online and offline data security has become a challenging issue, especially due to increase in the operational data. This research proposes a computational intelligent intrusion detection system using a Deep Neural Network (DNN). The dataset of University of South Wales NB15 (UNSW NB15) is used to simulate network traffic and malicious attacks. The data is preprocessed and normalized before inputting the DNN model. When applied to the DNN model, the refined data creates a learning model. Final step in this process involves the calculation of accuracy, and prediction rate. This step uses binary and multi-class classifications comparatively, to ensure that the resulting model is sustainable for intrusion detection and efficacy, and generates better results as compared to other models for Intrusion Detection.