Assistant Professor Erza Aminanto Receives Best Paper Award at ICISC 2024 for Advancing Explainable AI in Intrusion Detection
Dr. Erza Aminanto (left side) receiving best paper award at ICISC 2024
Monash University, Indonesia's Cybersecurity Program is proud to announce that one of our esteemed faculty members, Dr. Erza Aminanto, Assistant Professor and Course Coordinator Master of Cybersecurity program, has been awarded the Best Paper Award at the prestigious International Conference on Information Security and Cryptology (ICISC 2024), Lotte World Tower, Seoul, South Korea. The recognition highlights Monash University, Indonesia's ongoing commitment to cutting-edge research in the fields of cybersecurity and cryptography.
The award-winning paper, titled "Does Ordinality in Encoding Matter for XAI? A Case Study on Explainable IDS", was co-authored with Harry, a colleague from NICT (National Institute of Information and Communications Technology) Japan, and presented groundbreaking advancements in the interpretability of Intrusion Detection Systems (IDS) using Explainable Artificial Intelligence (XAI).
About the Paper
Intrusion Detection Systems are critical for securing computer networks against unauthorized access and malicious activities. However, the complexity of these systems, often powered by sophisticated machine learning models, creates challenges in terms of interpretability and transparency—a gap that this paper sought to bridge.
The research introduces Fusion SHAP, an innovative algorithm designed to enhance the interpretability of IDS models by addressing limitations in SHAP (SHapley Additive exPlanations) when handling one-hot encoded categorical features. Through this algorithm, the study delivers:
- Individual and merged explanations for one-hot encoded features.
- Comparative analysis with state-of-the-art methods, such as original SHAP and label encoding.
- Comprehensive insights into feature importance across seven types of attacks: DDoS, Backdoor, Injection, Password, Ransomware, Scanning, and XSS.
The experimental results demonstrated the effectiveness of Fusion SHAP in improving model transparency, contributing to increased trustworthiness and enabling more informed decision-making by security professionals.
ICISC 2024 and Its Significance
ICISC, recognized as a leading conference in the fields of information security and cryptography, covers topics such as AI security, applied cryptography, quantum cryptography, and side-channel attacks. This year, the conference had a highly competitive acceptance rate of approximately 20%, underscoring the quality and significance of the work presented.
A Remarkable Achievement
Reflecting on the honor, Assistant Professor Erza Aminanto expressed gratitude towards Monash University, Indonesia and collaborative researchers, emphasizing the importance of explainability in AI-driven cybersecurity solutions. "This award is not just recognition of our work but also a step forward in making complex cybersecurity systems more interpretable and effective," they said.
This accomplishment not only highlights the global impact of Monash University, Indonesia's research in cybersecurity but also showcases the university's dedication to fostering collaborations that address critical challenges in information security.
Congratulations to Assistant Professor Erza Aminanto and their co-author Harry for this remarkable achievement! Their work sets a new benchmark in explainable AI and network defense, reinforcing Monash University, Indonesia's position as a leader in innovative cybersecurity research.