摘要
人机对抗作为人工智能研究的前沿方向,已成为国内外智能领域研究的热点,并为探寻机器智能内在生长机制和关键技术验证提供有效试验环境和途径.本文针对巨复杂、高动态、不确定的强对抗环境对智能认知和决策带来的巨大挑战,分析了人机对抗智能技术研究现状,梳理了其内涵和机理,提出了以博弈学习为核心的人机对抗智能理论研究框架;并在此基础上论述了其关键模型:对抗空间表示与建模、态势评估与推理、策略生成与优化、行动协同与控制;为复杂认知与决策问题的可建模、可计算、可解释求解奠定了基础.最后,本文总结了当前应用现状并对未来发展方向进行了展望.
At the frontier of artificial intelligence research, human-computer gaming(HCG) technology has become a research hotspot. It provides an effective experimental environment and approach to exploring the intrinsic growth mechanism and verifying key technologies of machine intelligence. Considering the mounting challenges in intelligent decision-making posed by the complex, high-dynamic, and inconclusive environment coupled with strong confrontation, this paper analyzes the research status and dissects the key elements and intrinsic gaming mechanisms of HCG. This work also proposes a game learning-based theoretical research framework for HCG.Based on these analyses, we discussed HCG’s key models: gaming representation and modeling, situation assessment and reasoning, strategy generation, and optimization, as well as action coordination and control. The proposed research framework has laid a foundation for the modeling, computing, and interpreting solutions of complex cognition and decision problems. Finally, this paper summarizes the current application status and looks to future directions of development.
作者
黄凯奇
兴军亮
张俊格
倪晚成
徐博
Kaiqi HUANG;Junliang XING;Junge ZHANG;Wancheng NI;Bo XU(Center for Research on Intelligent System and Engineering,Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China)
出处
《中国科学：信息科学》
CSCD
北大核心
2020年第4期540-550,共11页
Scientia Sinica(Informationis)
关键词
人工智能
人机对抗
机器学习
智能博弈
认知决策
artificial intelligence
human-computer gaming
machine learning
autonomous intelligent gaming
command and decision making
