Workshops on Smart Manufacturing with Open and Scaled Data Sharing in Semiconductor and Microelectronics Manufacturing; Virtual and In-Person; Washington, DC; October/November 2023

半导体和微电子制造中开放和规模化数据共享的智能制造研讨会；

• 批准号: 2334590
• 负责人: James Davis
• 金额: $ 10万
• 依托单位: University of California-Los Angeles
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-08-15 至 2024-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2334590&HistoricalAwards=false
• 关键词: Workshops; Smart; Manufacturing; Open; Scaled

The Workshops on Smart Manufacturing with Open and Scaled Data Sharing in Semiconductor and Microelectronics Manufacturing will convene experts from the microelectronics manufacturing and artificial intelligence (AI) and machine learning (ML) research and practitioner communities, including representatives from relevant companies, universities, federal agencies, national laboratories, and the Manufacturing USA Institutes. The workshop is motivated by and addresses the critical need for manufacturing data in advanced manufacturing, a key finding of the previous workshop series, held under the auspices of the Subcommittee on Advanced Manufacturing and the Subcommittee on Machine Learning and Artificial Intelligence of the National Science and Technology Council. The inability to access manufacturing data that are contextualized, categorized, and discoverable was identified as a critical industry-wide impediment to smart manufacturing and the full potential of AI to the US economy. Manufacturing does not have the data use or delivery infrastructure for enhancing productivity that has evolved in many other industries. The aim of the workshops is to formulate a strategy for accessing cross-company microelectronics production data while keeping proprietary data secure. Such a strategy promises to improve the productivity of US microelectronics manufacturing companies by providing machine learning researchers with access to the well-characterized data. These data are needed to innovate new machine learning architectures specifically suited to categorizing manufacturing process data. The workshop is unique in its focus on detailed industry data the analysis of which defines critical business and technology decisions. The workshop is also unique in emphasizing the broader data science and computer science perspectives with a structure and format that facilitates building cross-community views and interests. The workshop will start with individual community discussions to share positions and understand vocabularies. These will grow into a series of several cross community virtual roundtables to engage manufacturing and AL/ML researchers with stakeholders from the microelectronics industry in defining specific business and technical questions. These outcomes will be addressed together in a subsequent in-person session in the Washington, DC area. The in-person session will: (1) define and agree on mechanisms for facilitating collaborative data sharing and model building at scale with current technologies and/or by defining needs for new methods, (2) frame an industry, academic, and government collaboration, and (3) create a draft action plan for cross-company data aggregation in the microelectronics manufacturing industry. The virtual and in-person sessions will include professional writers to assist the organizing committee in taking notes and rationalizing the findings from each session to guide the agenda for the remaining sessions. The in-person session will be followed by a virtual general review and comment session that involves all participants.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.

在半导体和微电子制造业中使用开放和规模的数据共享的智能制造研讨会将召集来自微电子制造和人工智能（AI）和机器学习（ML）研究和从业人员社区的专家，包括来自相关公司，联邦政府，大学，国家实验室以及美国制造商的代表。该研讨会是受到高级制造业制造数据的关键需求的动机，这是先前的研讨会系列的关键发现，该研讨会系列是由高级制造业小组委员会和国家科学技术委员会的机器学习和人工智能小组委员会举行的。无法访问上下文化，分类和可发现的制造数据被确定为对智能制造的关键障碍，以及AI对美国经济的全部潜力。制造业没有数据使用或交付基础架构来提高许多其他行业发展的生产率。研讨会的目的是制定访问跨公司微电子生产数据的策略，同时确保专有数据安全。这样的策略有望通过为机器学习研究人员提供访问良好的特征数据，从而提高美国微电子制造公司的生产率。需要这些数据来创新专门用于对制造过程数据进行分类的新机器学习体系结构。该研讨会在关注详细行业数据的关注点是独一无二的，该数据定义了关键业务和技术决策的分析。该研讨会在强调更广泛的数据科学和计算机科学观点方面也是独一无二的结构和格式，从而有助于建立跨社区的观点和兴趣。研讨会将从各个社区讨论开始，以分享职位并了解词汇。这些将成长为一系列跨社区虚拟圆桌会议，以使制造业和AL/ML研究人员与微电子行业的利益相关者一起定义特定的业务和技术问题。这些结果将在随后在华盛顿特区地区的面对面会议中共同解决。面对面的会议将：（1）定义并达成共识，以促进与当前技术和/或通过确定新方法的需求进行大规模合作数据共享和模型构建的机制，（2）构建行业，学术和政府协作，以及（3）制定跨公司数据集合草案在微电机制造行业中创建跨公司数据的草案。虚拟和面对面的会议将包括专业作家，以协助组织委员会做笔记并合理化每个会议的调查结果，以指导其余会议议程。面对面的会议将进行虚拟的一般审查和评论会议，该会议涉及所有参与者。该奖项反映了NSF的法定任务，并使用基金会的知识分子优点和更广泛的影响审查标准，被认为值得通过评估来获得支持。