Reinforcement Learning for Intrusion Detection: Recent Advances and Datasets

EasyChair Preprint 15665

Abstract

With the development of technologies and increasing security threats, intrusion detection systems have become more critical in detecting and protecting operations from attacks. Deep learning has significantly contributed to advancements in intrusion detection, especially through reinforcement learning systems. This survey reviews the concepts of intrusion detection and Reinforcement Learning (RL) systems in intrusion detection, with a focus on recent advancements using techniques such as Multi-Agent Reinforcement Learning (MARL), Adversarial Reinforcement Learning (AE-RL), and Inverse Reinforcement Learning (IRL). We also emphasize the crucial role of feature engineering in conjunction with RL techniques. By examining these cutting-edge approaches and their integration with advanced feature engineering methods, we aim to provide a comprehensive overview of the current state of the art in reinforcement learning-based intrusion detection systems and their potential to enhance cybersecurity measures. In addition to exploring the applications of reinforcement learning and feature engineering in intrusion detection, we highlight and analyze the most well-known databases used in this field, offering insights into the data resources that drive the development and evaluation of these advanced security systems.

Keyphrases: Intrusion Detection Dataset, Intrusion Detection System (IDS), Reinforcement Learning (RL), feature engineering

@booklet{EasyChair:15665,
  author    = {Dalal Fathi and Afraa Attiah and Abeer Hakeem and Asmaa Cherif},
  title     = {Reinforcement Learning for Intrusion Detection: Recent Advances and Datasets},
  howpublished = {EasyChair Preprint 15665},
  year      = {EasyChair, 2025}}