Adaptive policies for perimeter surveillance problems
European Journal of Operational Research(2020)
摘要
•Sequential search problem is formulated as a combinatorial multi-armed bandit.•We consider a novel setting with action-dependent filtering of Poisson rewards.•A new upper confidence bound approach based on martingale inequalities is proposed.•We derive order-optimal upper bounds on the regret of the approach.•We verify its strong performance and robustness in extensive numerical experiments.
更多查看译文
关键词
Applied probability,Stochastic processes,Uncertainty modelling,OR in defence
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要