摘要
深度学习作为人工智能领域的战略级技术,在2024年持续取得突破性进展,持续重构现代战争规则,成为大国军事智能化竞争的核心角力点。对2024年深度学习技术热门领域的主要发展进行了综合评述。研究了深度学习技术的发展现状,其中多模态推理技术、空间计算技术、鲁棒深度学习技术成为当前研究的热点方向;探讨了深度学习技术的军事应用前景,在战场侦察、态势感知、辅助决策、指挥控制等诸多军事领域任务中,深度学习技术的应用潜力较大并已经开始为实战赋能;分析了深度学习技术的发展趋势,多模态信号处理、模型压缩、小样本学习以及模型评测环境构建,将成为未来深度学习进一步赋能军事应用的重要发展方向。
Deep learning(DL),as a strategic-level technology in the field of artificial intelligence(AI),continues to make breakthrough progress in 2024,which continously reconstructs modern warfare rules,and becomes the core battleground for military intelligence competition among major powers.The major developments in the key fields of DL technology in 2024 are conducted.The current development status of DL is overviewed,among which multimodal reasoning,spatial computing and robust DL have become hot research directions.The military application prospects of DL are explored.Among many military tasks such as battlefield reconnaissance,situational awareness,decision-making assistance,command and control,the application potential of DL is significant and has begun to empower actual combat.The development trend of DL is analyzed.Multimodal signal processing,model compression,few-shot learning,and model evaluation environment construction will become important directions for further empowering military applications with DL in the future.
作者
相虎生
王亚珅
朱小伶
胡梦颖
常欢
刘小晗
欧阳小叶
Xiang Husheng;Wang Yashen;Zhu Xiaoling;Hu Mengying;Chang Huan;Liu Xiaohan;Ouyang Xiaoye(PAP Command College,Tianjin 300350,China;Artificial Intelligence Institute of CETC,Beijing 100049,China;CETC Academy of Electronics and Information Technology Group Co.,Ltd.,Beijing 100043,China;China Academy of Electronics and Information Technology,Beijing 100043,China)
出处
《战术导弹技术》
北大核心
2025年第2期64-74,共11页
Tactical Missile Technology
基金
军队装备技术基础项目(145BZB210005008X)。
关键词
深度学习
机器学习
人工智能
多模态推理
空间计算
鲁棒深度学习
deep learning
machine learning
artificial intelligence
multimodal reasoning
spatial computing
robust deep learning
