Guidelines and Resources for AI Model Access from TrusTEd Researchenvironments (GRAIMatter)

从 TrustTEd 研究环境访问 AI 模型的指南和资源 (GRAIMatter)

• 批准号: MC_PC_21033
• 负责人: Emily Jefferson
• 金额: $ 40.2万
• 依托单位: University of Dundee
• 依托单位国家: 英国
• 项目类别: Intramural
• 财政年份: 2022
• 资助国家: 英国
• 起止时间: 2022 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=MC_PC_21033
• 关键词: Guidelines; Resources; AI; Model; Access

Trusted Research Environments (TREs) provide a secure location for researchers to analyse data for projects in the public interest e.g. providing information to SAGE to fight the COVID-19 pandemic. TRE staff check outputs to prevent disclosure of individuals’ confidential data.TREs have historically supported only classical statistical data analysis. There is an increasing need to also facilitate the training of Artificial Intelligence (AI) models. AI has many valuable applications e.g., spotting human errors, streamlining processes, helping with repetitive tasks and supporting clinical decision making. The trained models then need to be exported from TREs for use. The size and complexity of AI models presents significant challenges for the disclosure-checking process. Models may be susceptible to external hacking: complicated methods to reverse engineer the learning process to find out about the data used for training, with more potential to lead to re-identification than conventional statistical methods.With input from public representatives, GRAIMatter will assess a range of tools and methods to support TREs to assess output from AI methods for potentially identifiable information, investigate the legal and ethical implications and controls, and produce a set of guidelines and recommendations to support all TREs with export controls of AI algorithms.

可信研究环境（TRE）为研究人员提供了一个安全的位置，以便为公共利益的项目分析数据，例如为SAGE提供抗击COVID-19大流行的信息。TRE工作人员检查输出，以防止个人的机密数据泄露。TRE历来只支持经典的统计数据分析。越来越需要促进人工智能（AI）模型的训练。AI有许多有价值的应用，例如，发现人为错误、简化流程、帮助完成重复性任务并支持临床决策。然后，需要从TREs导出经过训练的模型以供使用。人工智能模型的规模和复杂性给安全检查过程带来了重大挑战。模型可能容易受到外部黑客攻击：利用复杂的方法对学习过程进行逆向工程，以了解用于培训的数据，比传统的统计方法更有可能导致重新识别。在公众代表的参与下，GRAIMatter将评估一系列工具和方法，以支持TREs评估人工智能方法的输出，以获得潜在的可识别信息，调查法律的和道德的影响和控制，并制定一套指导方针和建议，以支持所有具有人工智能算法出口控制的TREs。

项目成果

Scottish Medical Imaging Service - Technical and Governance controls.

• DOI: 10.23889/ijpds.v7i3.1869
• 发表时间: 2022-08-25
• 期刊: International Journal of Population Data Science
• 作者: Caldwell J;Wallace R;Morris C;Fleming S;Baxter R;MacLeod R;Kerr W;Scobbie D;Rogers S;Ritchie F;Mansouri-Benssassi E;Krueger S;Jefferson E
• 通讯作者: Jefferson E

A National Network of Safe Havens: Scottish Perspective.

• DOI: 10.2196/31684
• 发表时间: 2022-03-09
• 期刊: Journal of medical Internet research
• 影响因子: 7.4
• 作者: Gao C;McGilchrist M;Mumtaz S;Hall C;Anderson LA;Zurowski J;Gordon S;Lumsden J;Munro V;Wozniak A;Sibley M;Banks C;Duncan C;Linksted P;Hume A;Stables CL;Mayor C;Caldwell J;Wilde K;Cole C;Jefferson E
• 通讯作者: Jefferson E

Machine learning models in trusted research environments -- understanding operational risks

可信研究环境中的机器学习模型——了解运营风险

• DOI: 10.23889/ijpds.v8i1.2165
• 发表时间: 2023
• 期刊: International Journal of Population Data Science
• 作者: Ritchie F

GRAIMATTER Green Paper: Recommendations for disclosure control of trained Machine Learning (ML) models from Trusted Research Environments (TREs)

• DOI: 10.5281/zenodo.7089491
• 发表时间: 2022-11
• 期刊: ArXiv
• 作者: E. Jefferson;J. Liley;Maeve Malone;S. Reel;Alba Crespi-Boixader;X. Kerasidou;Francesco Tava;Andrew McCarthy;R. Preen;Alberto Blanco-Justicia;Esma Mansouri-Benssassi;J. Domingo-Ferrer;J. Beggs;Antony Chuter;Christian Cole;F. Ritchie;A. Daly;Simon Rogers;Jim Q. Smith

Disclosure control of machine learning models from trusted research environments (TRE): New challenges and opportunities.

• DOI: 10.1016/j.heliyon.2023.e15143
• 发表时间: 2023-04
• 期刊: HELIYON
• 影响因子: 4
• 作者: Mansouri-Benssassi, Esma;Rogers, Simon;Reel, Smarti;Malone, Maeve;Smith, Jim;Ritchie, Felix;Jefferson, Emily
• 通讯作者: Jefferson, Emily