Aplicação de técnicas de inteligência artificial na otimização de sistemas de detecção de intrusão em redes de computadores

Abstract

Intrusion detection plays an important role in ensuring the security of information, it is the main technology in accurately identifying various attacks on the network. In this article, we explore how to model a network intrusion detection system based on deep learning and propose an approach to intrusion detection using recurrent neural networks (RNN-IDS). The experimental results show that the RNN-IDS is very suitable for modeling a classification model with high precision and that its performance is superior to that of traditional machine learning classification methods. In this article, a new IDS is proposed, which consists of a recurrent neural network with textit Gate Recurrent Unit (GRU) and textit Long Short Term Memory (LSTM). To carry out the experiments, the public database NSL-KDD was selected and the Dropout and K-fold regularization technique was used. The results proved that LSTM and GRU had better accuracy in relation to other classifiers, such as: J48, BayesNet, Randodm Forest, among others.