摘要
随着现代科技的不断革新,以机器学习尤其是深度学习为代表的人工智能技术正在改变无人系统的发展,推动无人作战等作战样式快速演变,对未来战争带来颠覆性影响。然而,由于深度学习的不可解释性、脆弱性等问题,人工智能技术在现实应用中产生了诸多不确定性和安全风险。本文聚焦人工智能技术在军事无人系统中的安全问题,从视觉感知的角度出发,重点分析了安全风险来源、对抗样本理论、视觉感知对抗攻击方法和防御对策,最后对无人系统领域人工智能应用的安全问题进行了总结。
With the rapid development of science and technology,artificial intelligence technology represented by machine learning,especially deep learning,is changing the development of unmanned systems,promoting new forms of the unmanned combat,and bringing about destructive impact to the future wars.However,due to the unaccountability and vulnerability of artificial intelligence,artificial intelligence technology has generated many uncertainty and security risks in real applications.This paper focuses on the security problems of artificial intelligence technology in military unmanned systems.From the perspective of visual perception,this paper focuses on analysis of the sources of security risks,adversarial example theory,visual adversarial attack and defense methods.Finally,the security issues of artificial intelligence applications in the field of unmanned systems are summarized.
作者
陈小前
姜廷松
姚雯
李超
CHEN Xiaoqian;JIANG Tingsong;YAO Wen;LI Chao(National Innovation Institute of Defense Technology,Academy of Military Science,Beijing 100071,China;School of Artificial Intelligence,Xidian University,Xi’an 710071,China)
关键词
人工智能安全
无人系统安全
视觉感知
对抗攻击与防御
artificial intelligence security
unmanned systems security
visual perception
adversarial attack and defense
