MRI: Acquisition of a GPU Cluster for Multi-Disciplinary Research and Education at University of Nevada, Las Vegas

该项目为购买和调试内华达大学拉斯维加斯分校(UNLV)的高性能图形处理单元(GPU)集群提供资金。该项目将解决跨多个学科的研究和教育工作中的紧急和长期需求、挑战和机遇：生物医学研究、智能交通系统和自动化车辆、基因组学、天文学和物理学。该项目将有助于推进大数据、战略计算、人工智能和智能基础设施系统方面的国家倡议。这些将整合、合成、建模和可视化来自各种来源的大量数据，并开发和应用人工智能(AI)技术来辅助决策。对于该项目的应用和基础研究方面，GPU集群将利用计算硬件、软件、传感器网络和通信系统方面的进步。该项目的一些内容将满足近期的社会需求，以保持和提高个人和家庭的生活质量，支持经济竞争力和企业增长，并促进社区的活力。这些主题包括与公共卫生、交通、环境和能源有关的主题。该项目的长期计划将致力于在这些领域以及天文学、物理学和基因组学的基础研究方面的探索和创新。这些活动将包括与学术界、政府实体和私营部门组织的伙伴关系。该项目将支持本科生和研究生的课程和联合课程活动，以增加他们对相关教育、研究和就业机会的兴趣。作为一个为少数族裔服务的机构和为拉美裔人服务的机构，这笔赠款将帮助联合国志愿人员组织为来自不同社会经济和社会人口社区的学生显著扩大这种机会。因此，该项目的一个结果将是帮助培养来自不同背景的熟练劳动力。GPU集群将支持基础和应用研究，以及跨多个学科的教育项目。研究工作的共同要素包括集成、合成、建模和可视化大量数据，以及各种人工智能技术的开发和应用，以支持决策。生物医学方面的努力将是使用可解释的深度学习技术，使用公开可用的多潜能转录因子数据集，对苯二氮类药物和阿片类药物过量的风险个体进行分层。与智能交通系统和自动化车辆相关的活动将解决交通网络上近实时应用的全面轨迹预测挑战，以帮助加快互联自动车辆和基础设施系统的部署；它们将使用来自各种车载、道路和路边传感器的数据。基因组学研究将更好地了解源于内源性逆转录病毒(ERV)整合的新的转录因子(TF)结合位点的进化。与天文学相关的努力将是使用卷积神经网络(CNN)从原行星盘图像估计行星质量。物理学方面的努力将是开发旋转等变CNN来模拟和评估原子尺度下的力场。该项目的教育方面将包括本科生和研究生层面的课程和联合课程倡议，以帮助提醒、参与、激发和激励学生在相关领域追求教育、研究和职业机会。该项目将包括与公共和私营部门组织和学术界的合作伙伴关系。该项目由重大研究仪器(MRI)计划、既定的激励竞争研究计划(EPSCoR)和计算机与信息科学与工程(CEISE)理事会共同资助。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。

The project funds the purchase and commissioning of a high-performance graphical processing unit (GPU) cluster at the University of Nevada, Las Vegas (UNLV). The project will address emergent and longer-term needs, challenges, and opportunities in research and educational efforts across multiple disciplines: biomedical research, intelligent transportation systems and automated vehicles, genomics, astronomy, and physics. The project will help advance national initiatives in big data, strategic computing, artificial intelligence, and smart infrastructure systems. These will integrate, synthesize, model, and visualize large volumes of data from various sources as well as develop and apply Artificial Intelligence (AI) techniques to assist decision making. For applied and basic research aspects of the project, the GPU cluster will leverage advances in computing hardware, software, sensor networks, and communications systems. Some elements of the project will address near-term societal needs to preserve and enhance the quality of living of individuals and families, support economic competitiveness and growth of businesses, and foster the vitality of communities. These include topics related to public health, transportation, environment, and energy. The project’s longer-term initiatives will address explorations and innovations in basic research in these domains as well as in astronomy, physics, and genomics. These activities will include partnerships with academia, government entities, and private sector organizations. The project will support curricular and co-curricular activities for undergraduate and graduate students to increase their interests in related education, research, and career opportunities. As a Minority-Serving Institution and Hispanic Serving Institution, this grant will help UNLV to significantly expand such opportunities for students from varied socio-economic and socio-demographic communities. Thus, an outcome of the project will be to help develop skilled work-forces from diverse backgrounds. The GPU cluster will support basic and applied research, as well as educational programs across multiple disciplines. Common elements for the research efforts include integrating, synthesizing, modeling, and visualizing large volumes of data along with the development and application of various AI techniques to support decision making. Efforts in Biomedicine will be to stratify individuals at risk for benzodiazepine and opioid overdose using interpretable deep learning techniques using publicly available pluripotency transcription factors datasets. Activities related to intelligent transportation systems and automated vehicles will address comprehensive trajectory prediction challenges for near real-time applications on transportation networks to help accelerate the deployment of Connected Automated Vehicles and Infrastructure Systems; they will use data from various in-vehicle, on-roadway, and roadside sensors. Research in Genomics will be to better understand the evolution of novel transcription factor (TF) binding sites originating from endogenous retrovirus (ERV) integration. Astronomy related endeavors will be to estimate planet mass from protoplanetary disk images using Convolutional Neural Networks (CNN). Efforts in Physics will be to develop rotationally equivariant CNN to simulate and evaluate force fields at atomistic scales of materials. Educational aspects of the project will include curricular and co-curricular initiatives at the undergraduate and graduate levels to help alert, engage, excite, and motivate students to pursue education, research, and career opportunities in related fields. The project will include partnerships with public and private sector organizations and academia.This project is jointly funded by the Major Research Instrumentation (MRI) program, the Established Program to Stimulate Competitive Research (EPSCoR), and the Computer & Information Science & Engineering (CISE) Directorate.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.