Enhanced intrusion detection with explainable ai

Accurate detection of network intrusions is essential for ensuring system security and protecting digital infrastructure. Traditional machine learning approaches often face limitations due to inadequate feature representation and lack of interpretability, resulting in reduced detection performance and limited trust in predictions. In this study, we propose an enhanced intrusion detection framework that integrates deep learning and ensemble learning techniques with Explainable AI (XAI). The model utilizes Convolutional Neural Networks (CNN) for effective feature extraction, while Random Forest and XGBoost are employed to improve classification accuracy and robustness. The system is designed to classify network traffic into multiple categories, including Benign, Combined Attack, DDoS, DoS GoldenEye, DoS Hulk, DoS Slowhttptest, DoS Slowloris, FTP-Patator, PortScan, and SSH-Patator. Furthermore, XAI techniques are incorporated to provide clear insights into model predictions, enhancing transparency and interpretability. This integrated approach ensures reliable intrusion detection while enabling better understanding of decision-making processes, contributing to improved cybersecurity management.

Keywords: Intrusion Detection System, CNN, Random Forest, XGBoost, Explainable AI (XAI), DDoS, DoS Attacks, PortScan, FTP-Patator, SSH-Patator, Network Security, Classification, Interpretability