arXiv · Multiagent Systems· Vahid Tavakkoli, Kabeh Mohsenzadegan, Kyandoghere Kyamakya·· 4 hr agoAI score25
SwarmReconGuard：黑盒检测由个体看似良性的智能体群体发起的分布式集体侦察
SwarmReconGuard: Black-Box Detection of Distributed Collective Reconnaissance by Individually Benign-Looking Agent Populations
AI brief
论文提出分布式集体侦察（DCR）威胁模型，并发布黑盒基准 SwarmReconGuard，防御方仅能观测服务边界遥测。Docker 隔离研究覆盖 11 种良性与攻击行为、10-10,000 个虚拟身份、440 次测试运行与 3,666,300 次请求。
Source: arXiv · Multiagent Systems · arxiv.org