Parallel Session: Next-Generation Combinatorial Fusion Analysis With Generative AI: A New Paradigm for Robust Intrusion Detection

Moderator: Frank Hsu, Ph.D., Clavius Distinguished Professor of Science, Professor of Computer & Information Science, Fordham University, New York

Cyberthreats such as Denial of Services (DoS) and Distributed Denial of Service (DDoS) present major challenges to the security of cyber infrastructures including interconnection networks, online services, and various Internet of Things (IoT’s). Although machine learning (ML) models have been used in intrusion detection system (IDS), they face challenges when detecting low-profile threats, using black box deep learning techniques, or relying on static models and fixed thresholds.  

In this talk, we propose the combinatorial fusion analysis (CFA) with generative AI to address and mitigate challenges in detecting low-profile and evolving threats in particular in imbalanced datasets. CFA considers both score combination and rank combination in Euclidean score and Kemeny rank space, respectively. In addition, it uses CD (cognitive diversity or rank-score diversity) between models to determine optimal combination of diverse and complementary models. This is a joint work with E. Owusu,  M. Mapkar, M. Rahouti, C. Schweikert, and D. C. Verma.