MRI: Development of an Instrument for Deep Learning Research

MRI：深度学习研究仪器的开发

• 批准号: 1725729
• 负责人: William Gropp
• 金额: $ 272.2万
• 依托单位: University of Illinois at Urbana-Champaign
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-10-01 至 2023-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1725729&HistoricalAwards=false
• 关键词: MRI; Development; Instrument; Deep; Learning

This project will develop and deploy a novel instrument for accelerating deep learning research at the University of Illinois (UI). The instrument will integrate the latest computing, storage, and interconnect technologies in a purpose-built shared-use system. This Instrument will deliver unprecedented performance levels for extreme data intensive emerging fields of research with far-reaching impacts in many areas, such as computer vision, natural language processing, artificial intelligence, healthcare and education. The instrument development will be driven by the UI deep learning (DL) community needs and will be carried out in collaboration with IBM and Nvidia. The instrument will serve as a focal point for the rapidly growing DL research community at UI, enable expansion of several research programs at UI, and contribute to STEM education and training.Specifically, the proposed instrument is a far-reaching cyberinfrastructure development for the research community and industry engaged with deep learning. The work will result in an advanced high-performing scalable instrument with capabilities far beyond those currently deployed in academia or industry to tackle large-scale deep learning projects. This instrument will serve as a focal point for a community-driven effort to advance the field of DL, integrating the work of computer scientists, systems engineers, and software developers. This project is transformative both in the systems architecture and domain science fields it will imbue, with new knowledge to be developed via new interactions and synergies that will emerge as part of this effort.The proposed development of this well-integrated instrument will improve the quality and expand the scope of research and training, provide inter- and intra-organizational use amongst many disciplines, and engage private sector partners. The work will have deep and long-lasting effects on future computer architectures for compute- and data-intensive applications by making the blueprints of the novel system architecture of the instrument publically available. Access to the high-performance software developed through this project will aid numerous science domains that utilize DL frameworks. The unprecedented computational capabilities of many applications will make it possible to tackle complex science, engineering and societal problems in many important fields ranging from education, to healthcare, to artificial intelligence (AI). This project will strive to include participants from under-represented minority and female students, to make new discoveries, train, and educate a new generation of users fluent with DL tools and methodologies, contributing to the development of a highly educated, and diverse workforce with specialized skillsets. Finally, the work will enable new industry-academic collaborations benefiting both the scientific community and industry nationwide.

该项目将开发和部署一种新的工具，用于加速伊利诺伊大学（UI）的深度学习研究。该仪器将在专用共享系统中集成最新的计算、存储和互连技术。该仪器将为极端数据密集型新兴研究领域提供前所未有的性能水平，并在许多领域产生深远影响，如计算机视觉，自然语言处理，人工智能，医疗保健和教育。仪器开发将由UI深度学习（DL）社区需求驱动，并将与IBM和Nvidia合作进行。该仪器将作为UI快速发展的DL研究社区的焦点，使UI的几个研究项目得以扩展，并有助于STEM教育和培训。具体而言，拟议的仪器是研究社区和从事深度学习的行业的深远的网络基础设施开发。这项工作将产生一种先进的高性能可扩展工具，其能力远远超出目前学术界或工业界部署的解决大规模深度学习项目的能力。该工具将作为社区驱动的努力推进DL领域的焦点，整合计算机科学家，系统工程师和软件开发人员的工作。该项目在系统架构和领域科学领域都具有变革性，它将注入通过新的互动和协同作用开发的新知识，这些互动和协同作用将成为这一努力的一部分。拟议开发的这一综合性很强的工具将提高研究和培训的质量并扩大其范围，在许多学科之间提供组织间和组织内的使用，并吸引私营部门合作伙伴。这项工作将对未来的计算机体系结构的计算和数据密集型应用产生深刻而持久的影响，使该仪器的新系统架构的蓝图在计算机上可用。访问通过该项目开发的高性能软件将有助于利用DL框架的许多科学领域。许多应用程序前所未有的计算能力将使其能够解决从教育到医疗保健到人工智能（AI）等许多重要领域的复杂科学，工程和社会问题。该项目将努力包括来自代表性不足的少数民族和女学生的参与者，以进行新的发现，培训和教育新一代熟练使用DL工具和方法的用户，为发展受过高等教育的多元化劳动力做出贡献。最后，这项工作将使新的产业-学术合作惠及全国科学界和产业界。

项目成果

Exploring HW/SW Co-Design for Video Analysis on CPU-FPGA Heterogeneous Systems

探索 CPU-FPGA 异构系统上视频分析的硬件/软件协同设计

• DOI: 10.1109/tcad.2021.3093398
• 发表时间: 2021
• 期刊: IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
• 影响因子: 2.9
• 作者: Zhang, Xiaofan;Ma, Yuan;Xiong, Jinjun;Hwu, Wen-mei;Kindratenko, Volodymyr;Chen, Deming
• 通讯作者: Chen, Deming

Unsupervised Discovery of Dynamic Neural Circuits

动态神经回路的无监督发现

• 发表时间: 2019
• 期刊: 33rd Conference on Neural Information Processing Systems
• 作者: Colin Graber, Ryan Loh

Graph Structured Prediction Energy Networks

• 发表时间: 2019-10
• 期刊: ArXiv
• 作者: Colin Graber;A. Schwing

tensorflow-tracing: A Performance Tuning Framework for Production

tensorflow-tracing：生产性能调优框架

• 发表时间: 2019
• 期刊: 2019 USENIX Conference on Operational Machine Learning (OpML ’19
• 作者: Hashemi, Sayed Hadi;Rausch, Paul;Rabe, Benjamin;Chou, Kuan-Yen;Liu, Simeng;Kindratenko, Volodymyr;Campbell, Roy H
• 通讯作者: Campbell, Roy H

Critical Minerals Map Feature Extraction Using Deep Learning

使用深度学习提取关键矿物图特征

• DOI: 10.1109/lgrs.2023.3310915
• 发表时间: 2023
• 期刊: IEEE Geoscience and Remote Sensing Letters
• 影响因子: 4.8
• 作者: Luo, Shirui;Saxton, Aaron;Bode, Albert;Mazumdar, Priyam;Kindratenko, Volodymyr
• 通讯作者: Kindratenko, Volodymyr