REU Site: Applying Data Science on Energy-efficient Cluster Systems and Applications

加州州立大学北岭分校将建立一个新的数据科学本科生研究经验（REU）网站。数据科学、并行和分布式计算技术已被广泛用于科学和工程中复杂的计算密集型和数据密集型问题的建模。为计算/数据密集型应用开发有效的节能技术在业务规划和决策中变得越来越重要。缺乏能够通过数据科学和并行和分布式计算技术综合利用各种测量方法、能耗模型和节能策略来降低能源成本的整体节能解决方案。该项目将通过应用数据科学和并行和分布式计算技术，为集群系统和各种应用开发节能解决方案。它将为学生提供身临其境的体验，他们将参加培训和研究活动，将数据科学和并行和分布式计算的知识和技能应用于节能集群系统和应用程序。学生将通过分析现有研究，设计和进行实验，并在数据收集，分析，处理和建模中应用数据科学技术来获得解决问题的研究经验。该项目的目标包括增加本科生的参与，特别是来自计算领域代表性不足的群体，参与数据科学和节能计算的研究;并应用数据科学构建节能集群系统和应用程序。该项目将招收8名本科生，每年暑期从事4个项目的工作，为期8周，共3年。在这个REU网站的24名参与学生中，超过60%将来自REU网站以外的大学/学院，超过60%将来自代表性不足的群体。在这个项目中，学生将培养智能数据收集，数据处理和地理空间数据和阴影地图的数据可视化技能;获得应用数据科学技术和方法来模拟集群系统和汽车空调系统的能耗的专业知识;并研究集群系统和各种应用程序上工作负载管理的节能解决方案（地理信息可视化、汽车空调系统、阴影检测与制图）。集群系统负载管理的节能解决方案和开发的应用程序可以作为指导，分析能源消耗数据，建立预测能源模型，并制定节能的集群系统和数据密集型应用程序的负载管理策略和解决方案。该项目中产生的软件工具、实验结果、模型和节能解决方案将通过出版物、专业演示和网络访问作为共享资源免费传播。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

California State University, Northridge will establish a new Research Experiences for Undergraduates (REU) site on data science. Data science and parallel and distributed computing techniques have been widely used for modeling complex compute-intensive and data-intensive problems in science and engineering. Developing an effective energy conservation technique for compute/data-intensive applications has become increasingly critical in business planning and decision-making. There is a lack of holistic energy-efficient solutions capable of reducing energy costs by comprehensively utilizing a variety of measurement methods, energy consumption models, and energy-saving strategies through data science and parallel and distributed computing technologies. This project will develop energy-efficient solutions for cluster systems and various applications by applying data science and parallel and distributed computing technologies. It will provide students with an immersive experience in which they will participate in trainings and research activities to apply knowledge and skills of data science and parallel and distributed computing on energy-efficient cluster systems and applications. Students will gain research experience on problem-solving through analyzing existing research, designing and conducting experiments, and applying data science technologies in data collection, analysis, processing, and modeling. The goals of this project include increasing the participation of undergraduate students, especially from groups that are underrepresented in computing, in research on data science and energy-efficient computing; and applying data science to build energy-efficient cluster systems and applications. This project will recruit 8 undergraduate students to work on 4 projects in the summer for 8 weeks each year for a total of 3 years. Among the 24 participant students in this REU site, more than 60% will be selected from universities/colleges outside the REU site, and more than 60% will be from underrepresented groups. In this project, students will develop skills in intelligent data collection, data processing, and data visualization of geospatial data and shade maps; gain expertise applying data science technologies and methods to model the energy consumption of cluster systems and automobile air conditioning systems; and investigate energy-efficient solutions for workload management on cluster system and various applications (geo-information visualization, automobile air conditioner system, and shading detection and mapping). The energy-efficient solutions for cluster systems workload management and the developed applications can serve as guidance for analyzing energy consumption data, building predictive energy models, and developing energy-saving workload management strategies for cluster systems and solutions for data-intensive applications. Software tools, experimental results, models and energy-saving solutions produced in the project will be freely disseminated as shared resources through publications, professional presentations, and web access.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.