当前位置:首页 >英文主页 >中英对照 > 报告详情

CSWP:2024关键AI系统的代理验证和验证:代理设计流程白皮书(英文版)(16页).pdf

上传人: Y**** 编号:176507 2024-09-29 16页 484.52KB

下载:

1、NIST Cybersecurity White Paper NIST CSWP 31 Proxy Validation and Verification for Critical AI Systems A Proxy Design Process Phillip Laplante Joanna DeFranco Rick Kuhn Jeff Voas Computer Security Division Information Technology Laboratory Mohamad Kassab Engineering Division Penn State University Thi

2、s publication is available free of charge from:https:/doi.org/10.6028/NIST.CSWP.31 September 26,2024 NIST CSWP 31 Proxy Validation and Verification September 26,2024 for Critical AI Systems Certain commercial entities,equipment,or materials may be identified in this document in order to describe an

3、experimental procedure or concept adequately.Such identification is not intended to imply recommendation or endorsement by the National Institute of Standards and Technology(NIST),nor is it intended to imply that the entities,materials,or equipment are necessarily the best available for the purpose.

4、NIST Technical Series Policies Copyright,Use,and Licensing Statements NIST Technical Series Publication Identifier Syntax Publication History Approved by the NIST Editorial Review Board on 2024-09-03 How to Cite this NIST Technical Series Publication:Laplante P,DeFranco J,Kuhn R,Voas J,Kassab M(2023

5、)Proxy Validation and Verification for Critical AI Systems:A Proxy Design Process.(National Institute of Standards and Technology,Gaithersburg,MD),NIST Cybersecurity White Paper(CSWP)NIST CSWP 31.https:/doi.org/10.6028/NIST.CSWP.31 Author ORCID iDs Phillip Laplante:0000-0002-0415-271X Joanna DeFranc

6、o:0000-0001-8966-5532 Rick Kuhn:0000-0003-0050-1596 Jeff Voas:0000-0003-1139-3690 Mohamad Kassab:0000-0002-3647-8511 Contact Information cswp-31-commentsnist.gov National Institute of Standards and Technology Attn:Computer Security Division,Information Technology Laboratory 100 Bureau Drive(Mail Sto

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
本文主要介绍了使用代理验证和验证关键人工智能系统(CAIS)的五阶段过程。该过程包括: 1. 评估CAIS风险级别:根据美国网络安全基础设施和国家安全局(CISA)定义的16个关键基础设施部门,评估CAIS的风险级别。 2. 系统评估以找到代理等效物:使用提出的CAIS分类法,将CAIS的特征与测试代理(即非关键原型或数字孪生)进行匹配,以评估测试代理的功能等效性。 3. 验证代理:通过创建和测试每个代理的使用和误用案例,对代理进行验证。 4. 创建误用案例并分类风险:创建误用案例,并根据风险级别对它们进行分类。 5. 测试误用案例:测试误用案例,以验证代理的有效性。 该方法的关键在于能够展示和测量“相似性”,这可以通过从转移学习中获取的统计和其他测量来实现。
如何在关键AI系统中使用代理验证和验证? 代理验证和验证过程包括哪些阶段? 如何评估关键AI系统的风险水平?
客服
商务合作
小程序
服务号
折叠