MRI: Acquisition of a Composable Platform as a Service Instrument for Deep Learning & Visualization (COMPaaS DLV)

MRI：获取可组合平台作为深度学习的服务工具

• 批准号: 1828265
• 负责人: Andrew Johnson
• 金额: $ 99.74万
• 依托单位: University of Illinois at Chicago
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-10-01 至 2022-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1828265&HistoricalAwards=false
• 关键词: MRI; Acquisition; Composable; Platform; Service

This project is about acquiring a much-in-demand Graphics Processing Unit (GPU)-based instrument, to develop a Service Instrument for Deep Learning and Visualization called "COMPaaS DLV: COMposable Platform". The project aims to complement available campus computing resources via the campus's research network. It will be able to access local and remote computing and storage facilities funded, including XSEDE, Blue Waters, Deep Learning Instrument and Chameleon. This critical instrumentation will provide a platform to pursue fundamental science and engineering research training in deep learning (data mining and data analytics, computer vision, natural language processing, artificial intelligence), visualization (simulation, rendering, visual analytics, video steaming, image processing), and a combination of deep learning and visualization (e.g., when data is so large that it cannot be easily visualized, then deep learning is used to extract features of interest to be visualized). The instrumentation also enables investigation and contribution to societal issues in disciplines such as anthropology, biology, cybersecurity, data-literacy, fraud detection, healthcare, manufacturing, urban sustainability, and cyber-physical systems (e.g., autonomous cars). Its design utilizes state-of-the-art computer architecture, known as composable architecture, in which the computer's components (traditional processor, GPU, storage, and networking) form a fluid pool of resources, such that different applications with different workflows can be run simultaneously, with each configuring the resources it requires almost instantaneously, at any time. Given composable infrastructure scalability and agility, it is more beneficial than traditional clouds and clusters that are rigid, overprovisioned, and expensive.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.

该项目旨在购买一种急需的基于图形处理单元（GPU）的仪器，以开发一种用于深度学习和可视化的服务仪器，称为“COMPaaS DLV：COMposable Platform”。 该项目旨在通过校园的研究网络补充可用的校园计算资源。它将能够访问资助的本地和远程计算和存储设施，包括 XSEDE、Blue Waters、深度学习仪器和 Chameleon。这一关键仪器将为深度学习（数据挖掘和数据分析、计算机视觉、自然语言处理、人工智能）、可视化（模拟、渲染、视觉分析、视频流、图像处理）方面的基础科学和工程研究培训提供一个平台。 ），以及深度学习和可视化的结合（例如，当数据太大而无法轻松可视化时，则使用深度学习来提取感兴趣的特征进行可视化）。该仪器还可以对人类学、生物学、网络安全、数据素养、欺诈检测、医疗保健、制造、城市可持续发展和网络物理系统（例如自动驾驶汽车）等学科的社会问题进行调查和贡献。其设计采用最先进的计算机架构，称为可组合架构，其中计算机组件（传统处理器、GPU、存储和网络）形成一个流动的资源池，这样具有不同工作流程的不同应用程序可以同时运行，每个项目几乎可以随时随时配置所需的资源。考虑到可组合基础设施的可扩展性和敏捷性，它比僵化、过度配置和昂贵的传统云和集群更有利。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力优点和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Modality-Classification of Microscopy Images Using Shallow Variants of Deep Networks

• DOI: 10.1109/bibm49941.2020.9313467
• 发表时间: 2020-12
• 期刊: 2020 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)
• 作者: J. Trabucco;Pengyuan Li;C. Arighi;H. Shatkay;G. Marai

Moving from Composable to Programmable

从可组合转向可编程

• DOI: 10.1109/ipdpsw55747.2022.00209
• 发表时间: 2022
• 期刊: 36th IEEE International Parallel & Distributed Processing Symposium Workshops (IPDPSW
• 作者: Chen, Zhongyi;Renambot, Luc;Long, Lance;Brown, Maxine;Johnson, Andrew E.
• 通讯作者: Johnson, Andrew E.

COMPaaS DLV: Composable Infrastructure for Deep Learning in an Academic Research Environment

COMPaaS DLV：学术研究环境中深度学习的可组合基础设施

• DOI: 10.1109/icnp.2019.8888070
• 发表时间: 2019
• 期刊: 2019 IEEE 27th International Conference on Network Protocols (ICNP
• 作者: Brown, Maxine;Renambot, Luc;Long, Lance;Bargo, Timothy;Johnson, Andrew E.
• 通讯作者: Johnson, Andrew E.

Data driven estimation of electric vehicle battery state-of-charge informed by automotive simulations and multi-physics modeling

• DOI: 10.1016/j.jpowsour.2020.229108
• 发表时间: 2021-01-31
• 期刊: JOURNAL OF POWER SOURCES
• 影响因子: 9.2
• 作者: Ragone, Marco;Yurkiv, Vitaliy;Mashayek, Farzad
• 通讯作者: Mashayek, Farzad

Modulated Graph Convolutional Network for 3D Human Pose Estimation

• DOI: 10.1109/iccv48922.2021.01128
• 发表时间: 2021-10
• 期刊: 2021 IEEE/CVF International Conference on Computer Vision (ICCV)
• 作者: Zhiming Zou;Wei Tang