Deep Learning based Intelligent Intrusion Detection System

Abstract

In present time to protect networks, systems, software and web-applications are challenging as well as necessary. As network infrastructures continue to expand and evolve, the risk of cyber threats and malicious activities also escalates. Ensuring network security in the face of evolving cyber threats requires effective Intrusion Detection Systems (IDS) capable of accurately identifying potential threats. Traditional IDS approaches often struggle to handle the increasing complexity and volume of network data. So noteworthy study is going on to detect attacks using Deep Learning (DL) and Machine Learning (ML) techniques. However, the high dimensionality and complexity of network traffic data make it challenging to extract meaningful information. This research contains latest DL techniques that can be used to develop effective IDS, challenges related to current IDS and proposes a novel approaches to enhance the performance of IDS by reducing the dimensionality of network data through the fusion of Autoencoders (AE) and Principal Component Analysis (PCA). For the purposes of assessing these methods, NSLKDD, UNSW15NB and CICIDS-2017 datasets are used. The proposed approaches have performed better compared to the existing AE+LSTM-based approach. The Wilcoxon signedrank test has been applied to confirm statistical significance of the results.