说明:收录25万 73个行业的国家标准 支持批量下载
IEC Technical ISO Specification ISO/IEC TS 8200 First edition Information technology 2024-04 Artificial intelligence Controllability of automated artificial intelligence systems Technologies de I'information-Intelligence artificielle- Controlabilite des systemes d'intelligence artificiels automatisés Reference number ISO/IEC TS8200:2024(en) ISO/IEC 2024 ISO/IEC TS 8200:2024(en) COPYRIGHT PROTECTEDDOCUMENT @ ISO/IEC2024 Al rights reserved.Unless otherwise specified,or required in the context of its implementation,no part of this publication may be reproduced or utilized otherwise in any form or by any means,electronic or mechanical,including photocopying,or posting on the internet or an intranet,without prior written permission.Permission can be requested from either ISO at the address below or ISO's member body in the country of the requester. ISO copyright office CP401:Ch.de Blandonnet 8 CH-1214 Vernier,Geneva Phone:+41227490111 Email:[email protected] Website:www.iso.org Published in Switzerland OISO/IEC2024-All rights reserved ii 8200:2024(en) ISO/ IEC TS Contents Page Foreword Introductio 1 Scope 2 Normative references 3 Terms and definitions 4 Abbreviations 5 Overview 5.1 Concept of controllability of an Asystem.... 5.2 System state ..... 5.3 System state transition 5.3.1 Target of system state transition 5.3.2 Criteria ofsystem state transition 5.3.3 Process ofsystem state transition 5.3.4 5.3.5 Side effects 5.4 Closed-loop and open -loop systems Characteristics of AIsystem controllability 6 6.1 Control over an AI system .. 6.2 Process of control 11 6.3 Control points 12 Span of control 6.4 13 6.5 Transfer of control ... 13 6.6 Engagement of control 15 6.7 Disengagement ofcontrol 6.8 Uncertainty during control transfer Cost ofcontrol 6.9 6.9.1 Consequences of contrd 6.9.2Cost estimation for a control 18 6.10 Cost ofcontrol transfer 18 6.10.1 Consequences of control transfer 18 6.10.2 Cost estimation for a control transfer 6.11 Collaborative control 7 Controllability of AIsystem : 19 19 7.1 Considerations ........ 7.2 Requirements on controllability of AIsystems .. 20 7.2.1 General requirements 20 Requirements on controllability ofcontinuous learning systems 7.2.2 7.3 Controllability levels of AI systems .... 21 Design and implementation of controllability of AIsystems 8 22 8.1 Principles 8.2 Inception stage . 23 Design stage 24 8.3 8.3.1 24 General 8.3.2 24 Approach aspect 25 8.3.3 Architecture aspect 8.3.4 Training data aspect 25 8.3.5 Risk management aspect 8.3.6 Safety-critical AIsystem design considerations 25 8.4 Suggestions for the development stage 25 Verification and validation of AI system controllability 9 .26 9. 1 Verification 26 OISO /IEC2024-All rights reserved ii

.pdf文档 ISO-IEC TS 12791 2024 Information technology - Artificial intelligence - Treatment of unwanted bias in classification and regression machine learning tasks

文档预览
中文文档 42 页 50 下载 1000 浏览 0 评论 309 收藏 3.0分
温馨提示:本文档共42页,可预览 3 页,如浏览全部内容或当前文档出现乱码,可开通会员下载原始文档
ISO-IEC TS 12791 2024 Information technology - Artificial intelligence - Treatment of unwanted bias in classification and regression machine learning tasks 第 1 页 ISO-IEC TS 12791 2024 Information technology - Artificial intelligence - Treatment of unwanted bias in classification and regression machine learning tasks 第 2 页 ISO-IEC TS 12791 2024 Information technology - Artificial intelligence - Treatment of unwanted bias in classification and regression machine learning tasks 第 3 页
下载文档到电脑,方便使用
本文档由 人生无常 于 2025-03-25 01:48:28上传分享
站内资源均来自网友分享或网络收集整理,若无意中侵犯到您的权利,敬请联系我们微信(点击查看客服),我们将及时删除相关资源。