Lightweight No-Regret Online Learning in Repeated Stackelberg Security Games
IEEE Control Systems Letters · 2026 · 10 · 193-198
DOI: 10.1109/lcsys.2026.3671784 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Control Systems Letters, 2026. Verfasst von Guanda Chen und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 30 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Guangdong Basic and Applied Basic Research Funding, in part by Guangdong Scientific Research Platform and Project Scheme.
Abstract (Originalsprache)
This work investigates the problem of learning no regret defender strategies in repeated Stackelberg Security Games (SSGs) against an unknown sequence of attackers. We propose an efficient algorithm that combines the Decomposed Optimal Bayesian Stackelberg Solver (DOBSS) with the online combinatorial optimization algorithm Follow the Perturbed Leader (FPL). Specifically, the optimal strategy is dynamically computed by solving a mixed integer linear programing problem where random perturbation is strategically integrated with historical observations of the attacker sequence. Our algorithm provably achieves no-regret with respect to the optimal hindsight strategy and remains resilient against adversarial attacker sequences, even when an adversary is aware of the defender’s algorithmic framework and intentionally selects sequences causing a higher regret. Additionally, we prove that our algorithm achieves exponential reductions in both space and time complexity compared with sub-procedure of baseline methods. Comprehensive empirical studies confirm that our algorithm outperforms baseline methods by achieving lower regret bound and reduced computational cost.
Förderung
- National Natural Science Foundation of China
- Guangdong Basic and Applied Basic Research Funding
- in part by Guangdong Scientific Research Platform and Project Scheme
- Youth Talent Support Program of Guangdong Association for Science and Technology
- Guangdong Provincial Key Lab of Integrated Communication, Sensing and Computation for Ubiquitous Internet of Things
Zitieren
APA
Chen, G., Ji, Y., & Han, S. (2026). Lightweight No-Regret Online Learning in Repeated Stackelberg Security Games. IEEE Control Systems Letters, 10, 193-198. https://doi.org/10.1109/lcsys.2026.3671784
Vancouver
Chen G, Ji Y, Han S. Lightweight No-Regret Online Learning in Repeated Stackelberg Security Games. IEEE Control Systems Letters. 2026;10:193-198. doi:10.1109/lcsys.2026.3671784
BibTeX
@article{Chen2026,
author = {Chen, Guanda and Ji, Yiding and Han, Shuo},
title = {Lightweight No-Regret Online Learning in Repeated Stackelberg Security Games},
journal = {IEEE Control Systems Letters},
year = {2026},
volume = {10},
pages = {193--198},
doi = {10.1109/lcsys.2026.3671784}
}
EUVIMED