CRII: RI: Strategic Interaction in Adversarial Settings with Information Hubs.

CRII：RI：对抗环境中与信息中心的战略互动。

• 批准号: 1850477
• 负责人: FEI FANG
• 金额: $ 17.48万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-03-15 至 2021-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1850477&HistoricalAwards=false
• 关键词: CRII; RI; Strategic; Interaction; Adversarial

To ensure security, public safety, and environmental sustainability, law enforcement and security agencies often operate with a very limited budget and resources to combat illegal activities such as attacks on critical infrastructure, urban crimes, poaching, and illegal logging. Advanced game-theoretic models and algorithms have been developed for analyzing threats in such scenarios that can be described as having attackers and defenders. These techniques have been effective in helping defenders allocate resources to defend against threats. Now, in a modern information environment dominated by social networks and increased connectivity in all aspects of life, defenders will need to consider the influence of the "information hubs" (IHs) in that environment that collect, process and disseminate information. IHs include all sorts of connected devices in the growing "Internet of Things", as well as the people in the social networks. This project aims to design new game theoretic models that account for the influence of IHs on attacker-defender interactions and develop methods to help the defender optimally select and allocate IHs for best defensive effect. This project will also develop methods to find the optimal allocation of defensive units given the presence of IHs so that the defender can achieve maximal benefit with minimal resources. Towards gaining a full understanding of the role of IHs in adversarial settings, the researchers will (1) develop new game-theoretic approaches that extend models such as Stackelberg security games to allow the IHs to communicate with the agents; (2) theoretically analyze the computational complexity of optimally selecting and allocating IHs to maximize the defender's expected utility in the game, and provide Mixed Integer Linear Programming (MILP)-based solution approach to compute the optimal selection and allocation, as well as approximation schemes and heuristic algorithms for large-scale problem instances; and (3) develop MILP-based algorithms to compute robust strategies against uncertainties in information collection, processing and dissemination. The researchers will evaluate the new algorithms both in simulation and for real-world conservation scenarios.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

为了确保安全、公共安全和环境可持续性，执法和安全机构通常在预算和资源非常有限的情况下开展工作，以打击非法活动，如对关键基础设施的袭击、城市犯罪、偷猎和非法伐木。先进的博弈理论模型和算法已经被开发出来，用于分析这种情况下的威胁，可以被描述为具有攻击者和防御者。这些技术有效地帮助人权维护者分配资源，抵御威胁。 现在，在一个以社交网络为主导的现代信息环境中，在生活的各个方面都有越来越多的联系，维权者需要考虑在这种环境中收集、处理和传播信息的“信息中心”的影响。 IH包括不断发展的“物联网”中的各种连接设备，以及社交网络中的人。 该项目旨在设计新的博弈论模型，解释IH对攻击者-防御者互动的影响，并开发方法来帮助防御者最佳地选择和分配IH，以获得最佳的防御效果。该项目还将开发方法，以找到最佳分配的防御单位的存在，使防御者可以实现最大的利益与最小的资源。 为了全面了解IH在对抗环境中的作用，研究人员将（1）开发新的博弈论方法，扩展Stackelberg安全游戏等模型，以允许IH与代理进行通信;（2）从理论上分析了最优选择和分配独立H以最大化防守者在比赛中的期望效用的计算复杂度，并提供基于混合线性规划（MILP）的求解方法来计算最优选择和分配，以及用于大规模问题实例的近似方案和启发式算法;以及（3）开发基于MILP的算法来计算针对信息收集、处理和分发中的不确定性的鲁棒策略。研究人员将在模拟和现实世界的保护场景中评估新算法。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Iterative Bounding MDPs: Learning Interpretable Policies via Non-Interpretable Methods

迭代边界 MDP：通过不可解释的方法学习可解释的策略

• 发表时间: 2021
• 期刊: Proceedings of the AAAI Conference on Artificial Intelligence
• 作者: Topin, Nicholay;Milani, Stephanie;Fang, Fei;Veloso, Manuela
• 通讯作者: Veloso, Manuela

Dual-Mandate Patrols: Multi-Armed Bandits for Green Security

• DOI: 10.1609/aaai.v35i17.17757
• 发表时间: 2020-09
• 作者: Lily Xu;Elizabeth Bondi-Kelly;Fei Fang;A. Perrault;Kai Wang;Milind Tambe

To Signal or Not To Signal: Exploiting Uncertain Real-Time Information in Signaling Games for Security and Sustainability

• DOI: 10.1609/aaai.v34i02.5493
• 发表时间: 2020-04
• 作者: Elizabeth Bondi-Kelly;Hoon Oh;Haifeng Xu;Fei Fang;B. Dilkina;Milind Tambe

A Robot’s Expressive Language Affects Human Strategy and Perceptions in a Competitive Game

机器人的表达语言会影响人类在竞争性游戏中的策略和感知

• DOI: 10.1109/ro-man46459.2019.8956412
• 发表时间: 2019
• 期刊: 2019 28th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN
• 作者: Roth, Aaron M.;Reig, Samantha;Bhatt, Umang;Shulgach, Jonathan;Amin, Tamara;Doryab, Afsaneh;Fang, Fei;Veloso, Manuela
• 通讯作者: Veloso, Manuela

Mitigating Manipulation in Peer Review via Randomized Reviewer Assignments

• 发表时间: 2020-06
• 期刊: ArXiv
• 作者: Steven Jecmen;Hanrui Zhang;Ryan Liu;Nihar B. Shah;Vincent Conitzer;Fei Fang