CAREER: Enabling Trustworthy Upgrades of Machine-Learning Intensive Cyber-Physical Systems

职业：实现机器学习密集型网络物理系统的可信升级

• 批准号: 2143351
• 负责人: Weiming Xiang
• 金额: $ 49.9万
• 依托单位: AUGUSTA UNIVERSITY RESEARCH INSTITUTE, INC.
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-06-01 至 2027-05-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2143351&HistoricalAwards=false
• 关键词: CAREER; Enabling; Trustworthy; Upgrades; Machine

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).Cyber-Physical Systems (CPS) sustainably benefit from software upgrades throughout their life cycles. However, as CPS become machine-learning-intensive due to rapidly increasing interactions between CPS and machine learning technologies, two major distinguishing factors associated with machine learning techniques raise significant safety concerns about CPS upgrades which play a critical role in enabling lifetime safety assurance. First, upgrades of machine learning components, which inherently result in system changes, come at significant safety risk for safety-critical CPS due to the vulnerabilities of machine learning techniques. Second, the traditional safe-by-verification upgrade framework, in which upgrades and verification have to be two separate procedures, is no longer valid for machine learning processes that update instantaneously during system operations. This project targets these unique challenges by developing scalable verification and monitoring methods for upgrades as well as safe upgrade procedures to enable trustworthy upgrades and achieve lifetime safety assurance in machine-learning-intensive CPS.This project will advance the state-of-the-art in the research of safety in machine-learning-intensive CPS from local time windows to global life cycles. With the expected research results, machine learning components in CPS can upgrade with desired safety assurance for lifetime safety purposes. In particular, this project will develop a novel scalable incremental verification framework as well as self-adaptive runtime monitoring methods for upgrades of machine-learning-intensive CPS. The proposed approach will also design safety-assured upgrade procedures by developing novel upgrade renewal procedures, safety-aware upgrades, and safety backup co-design methods. The project will develop an indoor vision-based autonomous vehicle testbed with upgradable neural networks for a variety of upgrade scenarios to perform rigorous evaluations. The integration of research and education plans will address CPS workforce shortage gaps, develop CPS curriculum, and design hands-on training for students. Activities such as engagement in K-12 STEM camps and collaboration with government and industry partners are also designed to inspire students early and promote public understanding of CPS, which aims to build a healthy and sustainable CPS workforce supply. The designed activities are uniquely positioned to attract members of underrepresented groups with a focus to enhance the diversity of the federal, state, and local CPS workforce.This proposal was funded under the NSF CPS CAREER program.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该奖项全部或部分由2021年美国救援计划法案（公法117-2）资助。网络物理系统（CPS）在其整个生命周期中可持续地受益于软件升级。然而，由于CPS和机器学习技术之间的交互迅速增加，CPS变得机器学习密集型，与机器学习技术相关的两个主要区别因素引起了对CPS升级的重大安全问题，CPS升级在实现终身安全保证方面发挥着关键作用。首先，由于机器学习技术的漏洞，机器学习组件的升级必然会导致系统更改，这对安全关键型CPS来说存在重大安全风险。其次，传统的安全验证升级框架，其中升级和验证必须是两个独立的程序，不再适用于在系统操作期间即时更新的机器学习过程。该项目针对这些独特的挑战，通过开发可扩展的升级验证和监控方法以及安全升级程序，实现机器学习密集型CPS的可靠升级，并实现终身安全保证。该项目将推动机器学习密集型CPS安全性研究的最新水平，从局部时间窗口到全局生命周期。有了预期的研究结果，CPS中的机器学习组件可以升级，以达到终身安全的目的。特别是，该项目将开发一个新的可扩展的增量验证框架以及自适应运行时监控方法，用于机器学习密集型CPS的升级。所提出的方法还将通过开发新的升级更新程序、安全意识升级和安全备份协同设计方法来设计安全保证升级程序。该项目将开发一个基于视觉的室内自动驾驶汽车测试平台，该平台具有可扩展的神经网络，可用于各种升级场景，以进行严格的评估。研究和教育计划的整合将解决CPS劳动力短缺的差距，开发CPS课程，并为学生设计实践培训。参与K-12 STEM营地以及与政府和行业合作伙伴合作等活动也旨在尽早激励学生并促进公众对CPS的了解，旨在建立健康和可持续的CPS劳动力供应。设计的活动是独特的定位，以吸引代表性不足的群体的成员，重点是提高联邦，州和地方CPS劳动力的多样性。这项建议是根据NSF CPS职业计划资助。这个奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

项目成果

Guaranteed approximation error estimation of neural networks and model modification

• DOI: 10.1016/j.neunet.2022.03.023
• 发表时间: 2022-03
• 期刊: Neural networks : the official journal of the International Neural Network Society
• 作者: Yejiang Yang;Tao Wang;Jefferson P. Woolard;Weiming Xiang

Computationally efficient neural hybrid automaton framework for learning complex dynamics

• DOI: 10.1016/j.neucom.2023.126879
• 发表时间: 2023-10
• 期刊: Neurocomputing
• 影响因子: 6
• 作者: Tao Wang;Yejiang Yang;Weiming Xiang

Maximum output discrepancy computation for convolutional neural network compression

• DOI: 10.1016/j.ins.2024.120367
• 发表时间: 2024-04
• 期刊: Inf. Sci.
• 作者: Zihao Mo;Weiming Xiang

Guaranteed Quantization Error Computation for Neural Network Model Compression

神经网络模型压缩的保证量化误差计算

• DOI: 10.1109/icit58465.2023.10143141
• 发表时间: 2023
• 期刊: 2023 IEEE International Conference on Industrial Technology (ICIT
• 作者: Cooke, Wesley;Mo, Zihao;Xiang, Weiming
• 通讯作者: Xiang, Weiming

Safety Verification of Neural Network Control Systems Using Guaranteed Neural Network Model Reduction

使用保证神经网络模型约简的神经网络控制系统的安全验证

• DOI: 10.1109/cdc51059.2022.9992984
• 发表时间: 2022
• 期刊: 2022 IEEE 61st Conference on Decision and Control (CDC
• 作者: Xiang, Weiming;Shao, Zhongzhu
• 通讯作者: Shao, Zhongzhu