Performance Analysis of Deep Neural Network and LSTM models for Secure Network Intrusion Detection System
R. Kavitha, S. Amutha · 2022
The Secure Network Intrusion Detection System (NIDS) is an important data security prevention appliance. To provide a high-performance Intrusion Detection System (IDS) can be extremely effective to prevent malicious behavior and cyber-attacks. However, Traditional machine learning methods are heavily reliant on manually created features. Deep learning is quite possibly the most exciting procedure utilized as of late by system for Detecting Network Intrusions to upgrade their exhibition in getting hosts and computer networks. In this study focuses on deep IDS approaches and investigates to achieve better results at different stages of intrusion detection process. In addition to that, analyze following deep learning algorithms such as deep neural network (DNN), long short term memory-Convolutional neural network (CNN-LSTM) and recurrent neural network (RNN) besides order to achieve the highest level of accuracy and precision with minimum number of iterations. For each model, concentrate on the exhibition in two classifications of characterization (binary and multi-class) under genuine traffic dataset NSL-KDD. The test results show that the RNN can classify attack types with an accuracy of 99.4%. In comparison, the CNN-LSTM model achieved 95.4 % accuracy, while the DNN model achieved 91.8 % accuracy.