An Intrusion Detection System for Network Security Using Recurrent Neural Network

Abstract

To maintain the security of vulnerable network is the most essential thing in network system; for network protection or to eliminate unauthorized access of internal as well as external connections, various architectures have been suggested. Various existing approaches has developed different approaches to detect suspicious attacks on victimized machines; nevertheless, an external user develops malicious behaviour and gains unauthorized access to victim machines via such a behaviour framework, referred to as malicious activity or Intruder. A variety of supervised machine algorithms and soft computing algorithms have been developed to distinguish events in real-time as well as synthetic network log data. On the benchmark data set, the NLSKDD most commonly used data set to identify the Intruder. In this paper, we suggest using machine learning algorithms to identify intruders. A signature detection and anomaly detection are two related techniques that have been suggested. In the experimental study, the Recurrent Neural Network (RNN) algorithm is demonstrated with different data sets, and the system’s output is demonstrated in a real-time network context.