Collaborative Research: CyberTraining: Pilot: Research Workforce Development for Deep Learning Systems in Advanced GPU Cyberinfrastructure

随着人工智能的最新进展，深度学习系统和应用已经成为多个跨学科领域的驱动力。虽然先进的图形处理器网络基础设施的快速改进在很大程度上支持了这一演变，但普遍缺乏将应用程序驱动的深度学习技术与利用图形处理器网络基础设施实施这种技术相结合的全面培训材料。为了填补这一空白，该项目开发了一个在线研讨会，其中包括一套由来自五个学科的六名教员提供的跨学科尖端培训课程。本次研讨会聚焦于基于GPU的深度学习系统和应用程序的最新创新，培养了新一代网络基础设施用户和贡献者的社区，他们可以使用、开发和改进先进的GPU网络基础设施，用于他们的深度学习研究。这类培训工作增进了深度学习和GPU网络基础设施工作人员的知识，并随后有助于解决重要的科学和社会问题，包括地理中的水文测绘、航空航天中的空间环境预报以及交通运输中的自动驾驶和交通监测。研讨会还将吸引来自代表性不足群体的学员，包括少数民族学生和来自农村地区的研究人员。该项目开发的跨学科研讨会旨在使参与者，包括本科生、研究生和研究人员，提高他们的多学科技能，扩大他们的学术研究组合，发展他们的远程协作能力，并显著增强他们的职业竞争力。为了实现这一目标，密集的研讨会包括1)一套实践讲座模块，为学员提供先进的GPU网络基础设施中全套深度学习系统的全面知识和技能；2)由学术和工业研究机构邀请的知名科学家举办的关于先进GPU网络基础设施和深度学习系统的前沿研究和应用的一系列讲座；以及3)将讲座中介绍的技术应用于实践的远程开放式跨学科协作项目。此外，还开发了一个交互式在线培训系统的原型，为受训者提供计算资源，并跟踪他们的学习进度，以便更有效和高效地开展培训活动。该项目预计将在深度学习系统和应用程序方面培养一支未来的研究队伍，并扩大在研究和教育中采用先进的GPU网络基础设施。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

With the recent advancements in artificial intelligence, deep learning systems and applications have become a driving force in multiple transdisciplinary domains. While this evolution has been largely supported by the rapid improvements in advanced GPU cyberinfrastructure, comprehensive training materials are generally absent that combine application-driven deep learning techniques with the implementation of such techniques using the GPU cyberinfrastructure. To fill in this gap, this project develops an online workshop that comprises of a set of interdisciplinary cutting-edge training sessions offered by six faculty members from five disciplines. With a focus on the latest innovations in GPU-based deep learning systems and applications, this workshop fosters a community of the next-generation cyberinfrastructure users and contributors, who can use, develop, and improve advanced GPU cyberinfrastructure for their deep learning research. Such training efforts enhance the knowledge of the deep learning and GPU cyberinfrastructure workforce, and subsequently contribute to the solutions of important scientific and societal problems, including hydrographic mapping in geography, space environment nowcasting in aerospace, and autonomous driving and traffic monitoring in transportation. The workshop will also attract trainees from underrepresented groups, including minority students and researchers from rural areas.The interdisciplinary workshop developed in this project aims at enabling participants, including undergraduate seniors, graduate students, and researchers, to improve their multidisciplinary skillsets, extend their academic research portfolios, develop their remote collaboration capacities, and significantly strengthen their career competitiveness. To achieve this goal, the intensive workshop includes 1) a set of hands-on lecture modules that provide trainees with comprehensive knowledge and skills on the full stack of deep learning systems in advanced GPU cyberinfrastructure, 2) a series of talks on the cutting-edge research in advanced GPU cyberinfrastructure and deep learning systems and application given by renowned scientists invited from academic and industrial research institutes, and 3) a remote open-ended interdisciplinary collaborative project of applying techniques introduced in lectures into practice. In addition, a prototype of an interactive online training system is developed to provide computing resources for the trainees and to track their learning progress, for more effective and efficient training activities. The project is expected to develop a future research workforce in deep learning systems and applications and to broaden the adoption of advanced GPU cyberinfrastructure in research and education.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.