Explainable Prediction of features contributing to Intrusion Detection using ML algorithms and LIME

Satish Kumar Karna¹*, Prakash Paudel¹, Ruby Saud¹ and Mohan Bhandari^2
1NCIT Lalitpur, Kathmandu, Nepal
²Samriddhi College Bhaktapur, Nepal
*Corresponding Author: Satish Kumar Karna, NCIT Lalitpur, Kathmandu, Nepal.

Published: August 30, 2023

DOI: 10.55162/MCET.05.160

View Pdf

Abstract  

Intrusion Detection System is software or hardware that checks a network for malicious activities. Each illegal activity is often recorded either centrally using a Security information and event management (SIEM) system or notified to an administration. This paper proposes an intrusion detection system using machine learning algorithms such as decision trees, random forests and explainable AI (XAI) using a real-world Software-Defined Networking (SDN) dataset. The evaluation includes various intrusion scenarios like network scanning, denial of service (DoS) attacks, and unauthorized access attempts. Random Forest exhibits the best performance, achieving an average training accuracy of 99.23%, while the decision tree achieves 98.78% accuracy. The result of this study contributes to the advancement of intrusion detection systems and fosters the development of resilient security solutions in the realm of SDN. The research highlights the importance of leveraging ML algorithms in effectively identifying and mitigating network intrusions, ultimately enhancing the security of SDN environments.

Keywords: Decision Tree; Intrusion Detection; Random Forest; XAI; LIME