Cost-Aware Network Disruption via Deep Reinforcement Learning

Dr Yang Lou, Hiroshima University, Japan

Abstract: Abstract: Network disruption seeks interventions that maximally degrade network functionality. Existing studies typically treat node- and edge-level attacks separately, overlooking differences in intervention costs. This talk presents a cost-aware disruption paradigm that unifies node and edge interventions within a single decision space, enabling adaptive selection of the most cost-effective target. To address this problem, a deep reinforcement learning based Cost-Aware Selector (CAS) framework is introduced to learn disruption policies over the unified intervention space. At each decision step, CAS identifies the intervention that maximizes disruption effectiveness relative to cost, leading to highly cost-efficient attack strategies. The proposed selection mechanism is generic and can be integrated with conventional and state-of-the-art learning-based disruption methods, resulting in a family of CAS-based attackers. Extensive experiments on synthetic and real-world networks demonstrate that CAS consistently outperforms both traditional and recent learning-based baselines in terms of cost-efficiency, highlighting the importance of jointly optimizing disruption effectiveness and intervention cost.

Speaker's Bio: Bio: Dr. Yang Lou received his Ph.D. degree from the Department of Electrical Engineering, City University of Hong Kong, Hong Kong, in 2017. He is currently an Associate Professor at the Graduate School of Advanced Science and Engineering, Hiroshima University, Japan. Prior to joining Hiroshima University, he held research and academic positions at City University of Hong Kong, Lingnan University, and the University of Osaka. From 2023 to 2025, he worked as an Associate Professor (awarded Youth Chair Professor) at the Department of Computer Science, National Yang Ming Chiao Tung University. He has published more than fifty research papers in prestigious IEEE Magazines and Transactions, such as CIM, CASM, TCYB, TNNLS, TNSE, and TCAS-I/II, as well as in renowned international conferences such as ICLR, GECCO, and IJCNN. He is a Senior Member of IEEE and a Fellow of the Higher Education Academy (FHEA). His research interests include network science and engineering, graph learning, machine learning, and optimization.