A novel feature selection with fuzzy deep neural network for attack detection in big data environment

Abstract

In recent times, the massive quantity of data and its continual expansion have transformed the significance of information security and data analysis systems for Big Data. An intrusion detection system (IDS) is commonly employed to monitor and analyze data for the detection of intrusions in the network. The conventional IDS models are not adequate to handle the high volume, variety, and speed of big data. This paper presents a new quantum brain storm optimization (QBSO) based feature selection with fuzzy deep neural network (FDNN), called QBSO-FDNN model for IDS in big data environment. The proposed model enables to detection of intrusions in the big data environment. The presented model initially performs preprocessing to enhance the quality of the big data. Also, to reduce the computational complexity, QBSO algorithm is applied to elect an optimal set of features. The choice of optimal features by the QBSO algorithm helps to boost the detection performance. Besides, FDNN model is applied as a classification model for identifying the occurrence of intrusions in the network. An extensive set of simulations was carried out to highlight the results on benchmark dataset. The resultant experimental values showcased the superior performance of the QBSO-FNN model with the detection accuracy of 98.90%.