A Hybrid Anomaly Based Intrusion Detection Methodology Using IWD for LSTM Classification

Abstract

The Network Intrusion Detection based on Anomaly is one of the best ways to identify the spam users and activities in cyber security. In present era, the Intrusion Detection System resources are increased due to inappropriate features that effect the detection rate of systems. To ensure better detection rate, a feature selection approach is utilized for the elimination of dissimilar and unemployable features in Intrusion Detection Systems. In addition, the time-consuming for the detection process also needs to be augmented for the process of classification. The paper introduces a method that avails the IWD algorithm for the feature subset selection in conjunction with LSTM to predict the malicious activity on that network. KDD CUP’99 dataset is employed for the judgement of performance on the intrusion detection in comparison with extant techniques. The performance estimate of the proposed model with previous methodologies depicts that the intended model is prominent by means of Higher Detection Rate, Low False Alarm Rate, and time consumption.