MULTI-OBJECTIVE WHALE OPTIMIZED WITH RECURRENT DEEP LEARNING FOR EFFICIENT INTRUSION DETECTION IN HIGH SENSITIVE NETWORK TRAFFIC

Intrusion detection plays a pivotal aspect in providing security for the information and the main technology lies in identifying different networks in an accurate as well as precise manner. By swift technological development, in recent years network systems are becoming highly susceptible to numerous revolutionary intrusion types. However, Deep learning based models are significant with accessible technique for detecting network intrusions of high network traffic. In this work, a method called, Multi-objective Whale Optimized with Recurrent Deep Learning (MWO-RDL) classifier for intrusion detection in high sensitive network traffic is proposed. With high sensitive network traffic as input, initially, Robust Scaler and Multi-objective Whale Optimization-based feature selection is designed for developing feature selection procedure as well as improving accuracy and finally the intrusion detection. The main idea behind the design of model is employed for assessing chosen feature subset that was explored in specified exploration space. Next, a classifier model called, Discrete Mutual Information-based Recurrent Deep Neural Learning is applied to the optimal selected features for classifying according to the characteristics of network traffic features into different type of attacks, normal traffic. Multi-objective Whale Optimized with Recurrent Deep Learning (MWO-RDL) is very suitable for modeling an intrusion detection model with high classification accuracy, intrusion detection rate and that its performance is comparatively better to that of traditional deep learning classification methods in multiclass classification. The MWO-RDL method minimizes false alarm rate of intrusion detection in a timely manner and bestows a novel research of high network traffic.