Modeling an intrusion detection using recurrent neural networks

Abstract

Cybercrime is one of the most difficult features of contemporary digital technology. Hence, it is crucial to restrict and perhaps even prevent its impacts. Systems are protected against numerous harmful attacks by Intrusion Detection (ID) systems. One method for figuring out a system’s typical behavior is to look at the call sequences the system processes make. In this paper, an ID system was devised based on the Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) architecture that is more accurate than conventional RNNs. The performance of the proposed ID model produced a high accuracy, detection rate, and low false alarm rate. Thus, supporting the model’s reputability.