论文标题
增强学习解决Stackelberg安全游戏的目标
Targets in Reinforcement Learning to solve Stackelberg Security Games
论文作者
论文摘要
强化学习（RL）算法已成功地应用于现实世界中的情况，例如非法走私，偷猎，森林砍伐，气候变化，机场安全等。这些场景可以用作Stackelberg Security Games（SSGS），在该场景中，被告和攻击者竞争目标资源竞争目标。该算法的能力可以通过哪个代理控制目标来评估。这篇综述研究了RL中SSG的建模，重点是RL算法中目标表示的可能改进。
Reinforcement Learning (RL) algorithms have been successfully applied to real world situations like illegal smuggling, poaching, deforestation, climate change, airport security, etc. These scenarios can be framed as Stackelberg security games (SSGs) where defenders and attackers compete to control target resources. The algorithm's competency is assessed by which agent is controlling the targets. This review investigates modeling of SSGs in RL with a focus on possible improvements of target representations in RL algorithms.
