摘要
为梳理深度学习技术在军事领域应用面临的难题,明确深度学习军事化应用攻关方向,首先从目标识别、态势感知和指挥决策等三方面总结了深度学习技术在军事领域的应用现状,然后分析了深度学习技术在军事领域应用所面临的难点与挑战。其在目标识别领域:面向稀缺认知样本的深度学习技术、不确定性信息条件下深度学习技术、实时性和基于无人平台的深度学习均有待突破。在态势感知领域:基于深度学习的战场态势大数据特征表示与挖掘技术、战场态势理解技术均有待突破。在指挥决策领域:深度学习的可解性有待提高,多实体协同决策技术、推理决策技术都有待提升。该研究成果能为深度学习技术在军事领域中创新发展与工程研究提供参考方向。
In order to sort out the difficulties in the application of deep learning technology in the military field and clarify the direction of deep learning in the application of public relations in the military field,this paper firstly summarizes the application status of deep learning technology in the military field from three aspects:target recognition,situation awareness and command decision-making,and then analyzes the difficulties and challenges in the application of deep learning technology in the military field.In the field of target recognition:deep learning technology for scarce cognitive samples,deep learning technology under uncertain information,its real-time and deep learning based on unmanned platform need to be broken through.In the field of situation awareness:the representation and mining technology of big data characteristics of battlefield situation based on deep learning,and the technology of battlefield situation understanding need to be broken through.In the field of command and decision-making:the solvability of deep learning needs to be improved,and the multi-entity collaborative decision-making technology and reasoning decision-making technology need to be improved.The research results can provide reference direction for the innovation and development of deep learning technology in the military field and engineering research.
作者
罗荣
王亮
肖玉杰
何翼
赵东峰
LUO Rong;WANG Liang;XIAO Yu-jie;HE Yi;ZHAO Dong-feng(Naval Research Academy,Beijing 102442;Beijing University of Posts and Telecommunications,Beijing 100876,China)
出处
《指挥控制与仿真》
2020年第1期1-5,共5页
Command Control & Simulation
基金
国防十三五预研项目资助
关键词
深度学习
人工智能
目标识别
态势感知
指挥决策
deep learning
artificial intelligence
target recognition
situation awareness
command decision