Abstract:The escalating Interconnectedness of devices has given rise to a surge in cyber-security threats, particularly zero-day attacks, which pose a significant challenge to traditional Intrusion Detection Systems (IDSs). This paper delves into the realm of behavior-based IDSs, leveraging Deep Neural Networks (DNNs), with a specific focus on the detection of network attacks. The effectiveness of these systems is closely tied to the quality of the training Dataset, where underrepresented samples can compromise detection performance, Our research centers on the development and evaluation of Deep Belief Networks (DBNs) as a robust solution for detecting cyber-attacks within networks of connected devices. The cornerstone of our investigation is the utilization of the CICIDS2017 Datasets, a comprehensive repository that facilitates both training and evaluation of our proposed DBN approach. To address the challenges posed by imbalanced Datasets, we employ various class balancing techniques, systematically evaluating their impact on detection performance. In the context of network attacks, our study pays special attention to notorious threats such as SQL injection, Cross-Site Scripting (XSS), Denial of Services (DoS), Distributed Denial of Services (DDoS), and other malicious activities. These attacks, often veiled in the subtleties of network traffic, demand sophisticated detection mechanisms. The empirical evaluation involves a comparative analysis with a conventional Multi-Layer Perceptron (MLP) model and the prevailing state-of-the-art IDSs. Our proposed DBN approach demonstrates competitive and promising results, showcasing significant performance improvements in the detection of attacks that are underrepresented in the training Datasets. The findings underscore the efficacy of DBNs in tackling the intricate challenges posed by emerging and sophisticated cyber threats. This research not only contributes to the advancement of intrusion detection methodologies but also provides insights into addressing the dynamic landscape of cyber threats, emphasizing the need for adaptive and intelligent systems in safeguarding networked environments.

A Deep Belief Networks Intrusion Detection Method Based on Generative Adversarial Networks

Detection Method of Domain Names Generated by DGAs Based on Semantic Representation and Deep Neural Network

Black-Box Adversarial Attacks Against Deep Learning Based Malware Binaries Detection with GAN

Ignnvd: A Novel Software Vulnerability Detection Model Based on Integrated Graph Neural Networks

An Intrusion Detection System Using a Deep Neural Network with Gated Recurrent Units.

An Adversarial Learning Model for Intrusion Detection in Real Complex Network Environments.

AEC_GAN: Unbalanced Data Processing Decision-Making in Network Attacks Based on ACGAN and Machine Learning

Network Intrusion Detection Using Deep Belief Network (DBN)

Enhancing Network Intrusion Detection Performance using Generative Adversarial Networks

Network intrusion detection based on IE-DBN model

An Enhanced AI-Based Network Intrusion Detection System Using Generative Adversarial Networks

Development of a Machine-Learning Intrusion Detection System and Testing of Its Performance Using a Generative Adversarial Network

Using EBGAN for Anomaly Intrusion Detection

IGAN-IDS: an Imbalanced Generative Adversarial Network Towards Intrusion Detection System in Ad-Hoc Networks

An Adaptive Enhancement Method of Malicious Traffic Samples Based on DCGAN-ResNet System

Adversarial Examples Detection of Electromagnetic Signal Based on GAN

Intrusion Detection Based on Generative Adversarial Network of Reinforcement Learning Strategy for Wireless Sensor Networks

DRL-GAN: A Hybrid Approach for Binary and Multiclass Network Intrusion Detection

Deep belief network based intrusion detection techniques: A survey

Deep Generative Learning Models for Cloud Intrusion Detection Systems

An efficient network behavior anomaly detection using a hybrid DBN-LSTM network