Intelligent FMI-Reduct Ensemble Frame Work for Network Intrusion Detection System (NIDS)

Abstract

The era of computer networks and information systems includes finance, transport, medicine, and education contains a lot of sensitive and confidential data. With the amount of confidential and sensitive data running over network applications are growing vulnerable to a variety of cyber threats. The manual monitoring of network connections and malicious activities is extremely difficult, leading to an increasing concern for malicious attacks on network-related systems. Network intrusion is an increasing issue in the virtual realm of the internet and computer networks that could harm the network structure in various ways, such as by altering system configurations and parameters. To address this issue, the creation of an efficient Network Intrusion Detection System (NID) that identifies malicious activities within a network has become necessary. The NID must regularly monitor network activities to detect malicious connections and help secure computer networks. The utilization of ML and mining of data approaches has proven to be beneficial in these types of scenarios. In this article, mutual a data-driven Fuzzy-Rough feature selection technique has been suggested to rank important features for the NIDS model to enforce cyber security attacks. The primary goal of the research is to classify potential attacks in high dimensional scenario, handling redundant and irrelevant features using proposed dimensionality reduction technique by combining Fuzzy and Rough set Theory techniques. The classical anomaly intrusion detection approaches that use individual classifiers Such as SVM, Decision Tree, Naive Bayes, k-Nearest Neighbor, and Multi Layer Perceptron are not enough to increase the effectiveness of detecting modern attacks. Hence, the suggested anomaly-based Network Intrusion Detection System named "FMI-Reduct based Ensemble Classifier" has been tested on highly imbalanced benchmark datasets, NSL_KDD and UNSW_NB15datasets of intrusion.