EAGER: A Data Flow Approach to Meet the Challenges of Big Data Analytics

EAGER：应对大数据分析挑战的数据流方法

• 批准号: 1649788
• 负责人: Lei Huang
• 金额: $ 30万
• 依托单位: Prairie View A & M University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-09-01 至 2021-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1649788&HistoricalAwards=false
• 关键词: EAGER; Data; Flow; Approach; Meet

The National Science Foundation uses the Early-concept Grants for Exploratory Research (EAGER) funding mechanism to support exploratory work in its early stages on untested, but potentially transformative, research ideas or approaches. This EAGER project was awarded as a result of the invitation in the Dear Colleague Letter NSF 16-080 to proposers from Historically Black Colleges and Universities to submit proposals that would strengthen research capacity of faculty at the institution. The project at Prairie View A & M University aims to implement machine learning algorithms on the data flow architecture and to conduct comprehensive performance and energy consumption studies in comparison with those of classical von Neumann computer architectures. The project outcomes can address the challenges modern computers based on the von Neumann architecture are facing to pertaining memory and power walls in the big data era. The project is the first attempt to implement machine-learning algorithms on the data flow architecture. The results of the project will demonstrate if the data flow model can be used successfully to meet the performance, energy efficiency, and scalability requirements of widely used big data analytics and machine learning applications. Once demonstrated, the work can lead to the design and implementation of better and faster high performance computers and Big Data Analytics applications. In addition, the project will also increase the research capability at Historically Black Colleges and Universities in High Performance Computing, Computer Architecture, and Big Data Analytics.This EAGER project is funded by the Directorate for Computer and Information Science and Engineering.

美国国家科学基金会使用探索性研究早期概念赠款（EAGER）资助机制，以支持未经测试但可能具有变革性的研究想法或方法的早期阶段的探索性工作。EAGER项目是由于在亲爱的同事信NSF 16-080中邀请来自历史上黑人学院和大学的提议者提交提案，以加强该机构教师的研究能力而获得的。Prairie View A M大学的项目旨在在数据流架构上实现机器学习算法，并与经典的von Neumann计算机架构进行全面的性能和能耗研究。该项目的成果可以解决基于冯诺依曼架构的现代计算机在大数据时代面临的内存和电源墙的挑战。该项目是在数据流架构上实现机器学习算法的第一次尝试。该项目的结果将证明数据流模型是否可以成功地用于满足广泛使用的大数据分析和机器学习应用程序的性能，能效和可扩展性要求。一旦被证明，这项工作可以导致更好，更快的高性能计算机和大数据分析应用程序的设计和实施。此外，该项目还将提高历史上黑人学院和大学在高性能计算、计算机体系结构和大数据分析方面的研究能力。EAGER项目由计算机和信息科学与工程局资助。

项目成果

Implementing Deep Neural Networks on Fresh Breeze

在 Fresh Breeze 上实施深度神经网络

• DOI: 10.3233/978-1-61499-843-3-539
• 发表时间: 2018
• 期刊: Advances in parallel computing
• 作者: Dennis, Jack B.;Huang, Lei;Lim, Willie;Wu, Hsiang-Huang;Yan, Yuzhong
• 通讯作者: Yan, Yuzhong