《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 人工智能 > 設(shè)計應(yīng)用 > 人工智能基礎(chǔ)軟硬件架構(gòu)的關(guān)鍵技術(shù)研究
人工智能基礎(chǔ)軟硬件架構(gòu)的關(guān)鍵技術(shù)研究
2020年信息技術(shù)與網(wǎng)絡(luò)安全第3期
楊建,,傅明駒,,徐國強
(江蘇自動化研究所 計算機事業(yè)部,,江蘇 連云港 222000)
摘要: 針對當前人工智能技術(shù)在軍用領(lǐng)域的應(yīng)用問題和難點,,研究基于深度學習技術(shù)的軍用軟硬件架構(gòu)設(shè)計方法,,并且結(jié)合業(yè)內(nèi)主流的軟硬件基礎(chǔ)架構(gòu),,分析在軍事應(yīng)用中涉及的關(guān)鍵技術(shù)和解決思路。通過對深度學習模型壓縮轉(zhuǎn)化,、數(shù)據(jù)增強和分布式訓(xùn)練等軟硬件架構(gòu)的關(guān)鍵技術(shù)研究,,設(shè)計構(gòu)建了一站式人工智能開發(fā)平臺,、嵌入式邊緣智能計算平臺,,對軍用人工智能提供端到端的訓(xùn)練、部署和測試支撐,。該研究可為未來智能無人作戰(zhàn)系統(tǒng)提供軟硬協(xié)同的智能計算解決方案。
中圖分類號:TP391.7
文獻標識碼:A
DOI: 10.19358/j.issn.2096-5133.2020.03.009
引用格式:楊建,,傅明駒,,徐國強.人工智能基礎(chǔ)軟硬件架構(gòu)的關(guān)鍵技術(shù)研究[J].信息技術(shù)與網(wǎng)絡(luò)安全,2020,39(3):44-49.
Research on key technologies of basic hardware and software architecture of artificial intelligence
Yang Jian,F(xiàn)u Mingju,,Xu Guoqiang
(BU of Computer,Jiangsu Automation Research Institute,Lianyungang 222000,China)
Abstract: Aiming at the application problems and difficulties of artificial intelligence technology in the military field,the design method of military hardware and software architecture based on deep learning technology is studied.Combined with the mainstream software and hardware infrastructure in the industry,the key technologies and solutions in military applications are analyzed. Through studying the key technologies of software and hardware architecture,including deep learning model compression transformation,data enhancement and distributed training,this paper designs and builds a onestop artificial intelligence development platform and an embedded edge intelligent computing platform to provide end to end training,department and test support for military AI.It provides a softwarehardware collaborative intelligent computing solution for future intelligent unmanned combat system.
Key words : military artificial intelligence;model compression;data enhancement;onestop artificial intelligence development platform;embedded edge intelligent computing platform

0    引言

未來戰(zhàn)爭中,,戰(zhàn)場認知與決策速度將成為戰(zhàn)爭勝負的決定因素,。誰能夠更快地處理戰(zhàn)場信息(感知)、理解態(tài)勢(認知),、判斷決策(決策),、執(zhí)行打擊(行動)等任務(wù),,誰就能壓縮OODA任務(wù)周期,從而在對戰(zhàn)中形成與對手的不對稱優(yōu)勢,。而人工智能在軍用領(lǐng)域的研究當前還處于初級階段,,屬于弱人工智能范疇。其對軍用信息系統(tǒng)的智能化分析與輔助能力基本停留在事務(wù)處理和業(yè)務(wù)計算層面,,對所處理的信息(如戰(zhàn)場數(shù)據(jù)、作戰(zhàn)分析,、態(tài)勢圖生成等)無法實現(xiàn)智能認知以及智能輔助決策,,無法滿足軍用OODA的需求,。此外,,在實際軍事應(yīng)用中要想取得對抗優(yōu)勢,,必須具備強大的計算能力支撐。因此,,需要研究人工智能在軍用領(lǐng)域中基礎(chǔ)軟硬件架構(gòu)所涉及的關(guān)鍵技術(shù),這對解決人工智能在軍用領(lǐng)域的落地問題和降低人工智能的應(yīng)用門檻具有重要的作用,。




本文詳細內(nèi)容請下載:http://forexkbc.com/resource/share/2000003179





作者信息:

楊建,,傅明駒,,徐國強

(江蘇自動化研究所 計算機事業(yè)部,,江蘇 連云港 222000)


此內(nèi)容為AET網(wǎng)站原創(chuàng),,未經(jīng)授權(quán)禁止轉(zhuǎn)載,。