A-IDS: An Adaptive Intrusion Detection System Using Reinforcement Learning-Based Deception

Abstract

Traditional rule-based intrusion detection systems can't keep up with the fast pace of today's cyber threats. In response, this paper introduces A-IDS, an AI-driven system that doesn't just spot attacks, it also responds to them automatically and intelligently. By blending machine learning to detect threats accurately with reinforcement learning to decide the best actions, A-IDS uses data augmentation to better handle rare and hard-to-detect attacks. Our proposed model was tested on the UNSW-NB15 dataset and demonstrated in a live deployment.