Associate Professor Madhusanka Liyanag, University College Dublin, Ireland

Topic - The AI Security Paradox: Protecting Networks with AI and Protecting AI in Networks

Abstract

Artificial Intelligence (AI) and machine learning (ML) are rapidly transforming the landscape of modern telecommunications and network management. Networks today face an unprecedented scale of complexity, with billions of connected devices, dynamic traffic patterns, and sophisticated adversarial threats. At the same time, AI-driven systems are increasingly embedded in critical network infrastructure, making the security of AI systems themselves a matter of paramount importance. This tutorial addresses the dual relationship between AI and network security: how AI can be leveraged as a powerful tool to strengthen network security, and how networks must be secured to protect the AI systems that operate within them. The first dimension of this tutorial, AI for Network Security, examines how AI and ML techniques can be applied to detect, prevent, and respond to network threats. Traditional rule-based and signature-driven security tools are increasingly outpaced by the volume, velocity, and variety of modern cyberattacks. AI-powered approaches, including deep learning-based intrusion detection, anomaly detection, federated learning for privacy-preserving threat intelligence sharing, and reinforcement learning for adaptive cyber defence, offer a new generation of tools capable of operating at network scale in real time. We will explore concrete use cases spanning intrusion detection systems (IDS), malware classification, network traffic analysis, and zero-trust architectures. The second dimension, Network Security for AI, addresses the growing recognition that AI systems deployed in network environments are themselves vulnerable to a wide range of attacks. Adversarial machine learning, data poisoning, model inversion, and model stealing are just a few of the threat vectors that can compromise AI-driven network functions. As AI becomes the backbone of network automation, intelligent resource management, and autonomous network operations (e.g., in 5G, 6G, and Open RAN), ensuring the integrity, robustness, and trustworthiness of AI models is critical. This portion of the tutorial will discuss threat taxonomies specific to AI systems in networking contexts, as well as defences including adversarial training, differential privacy, robust aggregation, and explainable AI for anomaly accountability.

See here for more details.

Biography

Dr. Madhusanka Liyanage is an Associate Professor/Ad Astra Fellow and Director of Graduate Research at the School of Computer Science, University College Dublin, Ireland. He is leading Network Softwarization and Security Labs (NetsLab) at the UCD School of Computer Science, a dynamic research group leading the charge in enhancing the security and privacy of next-generation mobile networks, including 5G and 6G. He is also an adjunct professor at the University of Oulu, Finland, the University of Ruhuna, Sri Lanka, and the University of Sri Jayawardhanepura, Sri Lanka. He received his Doctor of Technology degree in communication engineering from the University of Oulu, Oulu, Finland 2016. He also received the prestigious Marie Skłodowska-Curie Actions Individual Fellowship and the Government of Ireland Postdoctoral Fellowship during 2018-2020. In 2020, he received the “2020 IEEE ComSoc Outstanding Young Researcher” award by IEEE ComSoc EMEA. In 2021,2022, 2023 and 2024, he was ranked among the World’s Top 2% Scientists (2020, 2021, 2022 and 2023) in the List prepared by Elsevier BV, Stanford University, USA. Also, he was awarded an Irish Research Council (IRC) Research Ally Prize as part of the IRC Researcher of the Year 2021 awards for his positive impact as a supervisor. In 2022, he received “the 2022 Tom Brazil Excellence in Research Award” from the SFI CONNECT Center. Moreover, Madhusanka received a Special Commendation for IRC Early Career Researcher of 2022 by the Irish Research Council, Ireland. Dr Liyanage’s research interests are 5G/6G Security, Blockchain, Artificial Intelligence (AI), Explainable AI and Federated Learning (FL) security, Network Slicing, Internet of Things (IoT) and Multi-access Edge Computing (MEC). He has co-authored over 250 publications, including three authored books, four edited books and several patents in the mobile network security domain. He is also an expert consultant at the European Union Agency for Cybersecurity (ENISA) and a Funded Investigator of the Science Foundation Ireland CONNECT Research Centre, Ireland. Moreover, he is an expert reviewer for different funding agencies in Europe, Asia, and Oceania.