Planning Grant: Engineering Research Center for Edge Intelligence

工程研究中心（ERC）的规划赠款竞争是作为ERC计划内的试点招标运行。 规划赠款不需要作为整个ERC竞赛的一部分，但旨在建立团队之间的能力，以规划融合的，中心规模的工程研究。该补助金将为拟议的边缘智能中心提供规划资源。该中心将开发由物联网（IoT）系统驱动的数据科学新技术。物联网系统与真实的世界交互：工业、医疗保健、农业等。物联网系统通过收集和处理来自真实的世界的数据，为这些和其他应用领域带来了巨大的好处。然而，需要新的方法来利用物联网系统的潜力。物联网设备必须在电力和通信方面受到限制。这些因素意味着这些智能系统的关键方面必须体现在大型数据中心之外，在互联网的边缘，靠近数据创建和使用的点。这项规划赠款有三个目标。 规划拨款工作将确定一个完整的中心团队，包括成员机构研究人员，多样性和包容性协调员，劳动力发展协调员和创业协调员。提案团队将在研究人员、工业合作伙伴、企业家和教育工作者中确定广泛的利益相关者。整个中心团队和利益相关者将为边缘智能中心制定战略计划。边缘智能是三个领域的融合：数据科学，低功耗计算和分布式系统。数据科学和物联网系统通常由非常不同的团队研究、开发和部署。边缘智能中心的使命将是建立一个跨行业和学术界的利益相关者社区，以开发和利用这一新兴技术的能力。该中心的智能方法将基于几个基础：机器学习，统计，低功耗计算，分布式算法。物联网系统广泛应用于制造业、医疗保健、物流、农业和许多其他领域。应用于物联网系统的机器学习可以提供新水平的定制和改进的系统操作。边缘智能中心将在基础、工程和应用方面采用融合的整体研究方法。基础目标包括为带宽和功率有限的分布式计算系统设计的机器学习算法。增量学习算法，用于为特定环境定制和更新系统。工程目标包括在分布式平台上执行机器学习的分布式系统，包括边缘节点、雾中心和云。低功耗机器学习系统基于边缘和中心的增量训练的低功耗方法。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The Planning Grants for Engineering Research Centers (ERC) competition was run as a pilot solicitation within the ERC program. Planning grants are not required as part of the full ERC competition, but intended to build capacity among teams to plan for convergent, center-scale engineering research.This grant will provide planning resources for a proposed Center for Edge Intelligence. The proposed center will develop new techniques for data science driven by Internet-of-Things (IoT) systems. IoT systems interact with the real world: industry, health care, agriculture, etc. IoT systems promise substantial benefits for these and other application areas by gathering and processing data from the real world. However, new approaches are required to harness the potential of IoT systems. IoT devices must operate with limits on both electrical power and communication. These factors mean that key aspects of these intelligent systems must be embodied outside of large data centers, at the edge of the Internet, close to the points at which data are created and used. This planning grant has three goals. The planning grant effort will identify a full center team including member institution researchers, diversity and inclusion coordinators, workforce development coordinators, and entrepreneurship coordinators. The proposing team will identify a broad set of stakeholders among researchers, industrial partners, entrepreneurs, and educators. The full center team and stakeholders will create a strategic plan for the Center for Edge Intelligence.Edge Intelligence is the convergence of three fields: data science, low-power computing, and distributed systems. Data science and IoT systems are typically studied, developed, and deployed by very different groups. The Center for Edge Intelligence's mission will be to build a community of stakeholders across industry and academia to develop and harness the capabilities of this emerging technology. The intellectual approach of the Center will be based on several foundations: machine learning, statistics, low-power computing, distributed algorithms. Internet-of-Things systems are widely used in manufacturing, health care, logistics, agriculture, and many other areas. Machine learning applied to IoT systems can provide new levels of customization and improved system operation. The Center for Edge Intelligence will pursue a convergent, holistic research approach in foundations, engineering, and applications. Foundational goals include machine learning algorithms designed for distributed computing systems with limited bandwidth and power. Incremental learning algorithms to customize and update systems for their particular environment. Engineering goals include distributed systems that perform machine learning on distributed platforms, including edge nodes, fog hubs, and the cloud. Low-power machine learning systems. Low-power approaches to edge- and hub-based incremental training. Applications include manufacturing, health care and wellness, agriculture, transportation and logistics.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.