NSF Workshop on Real-time Learning and Decision Making of Dynamical Systems. To Be Held at NSF, February 12-13, 2018.

研讨会的目标是拥有一组具有互补背景的领先专家（在控制，信号处理，机器学习，通信，电力和能源，运输等领域）。跨越各个研究领域之间的桥梁，并塑造从数据驱动的动态系统的实时学习中产生的研究范式。具体来说，我们已经确定了以下数据丰富的动态工程系统：电力和能源系统，交通系统，以及信号和信息处理系统。讨论的主题包括工程系统的各种真实的时间学习方法（如深度学习架构，基于模型的学习，无模型学习，强化学习等），工程系统的数据表示（包括特征提取、图形模型、真实的实时无监督学习等方面的研究问题），以及关闭数据循环的潜在解决方案。该研讨会将研究控制，信号处理，机器学习，通信，电力和能源以及运输系统。智力优势：该研讨会将解决动态数据驱动工程系统中实时学习和决策的关键问题。感兴趣的主题包括从控制，信号处理，机器学习，计算智能和领域应用的观点学习和决策的不同观点。他们将协同努力，为NSF的十大理念中的两个做出贡献：“利用数据促进21世纪世纪科学和工程”和“在NSF发展融合研究”。更广泛的影响：本次研讨会将参与并促进以工程系统中的实时决策和数据为中心的融合研究。这些讨论将对机器学习和大规模数据驱动工程系统的研究和教育产生重大影响。讲习班还将为学术界和工业界之间的合作研究创造机会。还将为可能的创新和创业产生想法。研讨会报告将成为科学界和公众的重要资源。

The goal of the workshop is to have a group of leading experts that have complementary background (in the area of control, signal processing, machine learning, communication, power and energy, transportation, etc.) to cross the bridge among various research areas and shape the research paradigm that arises from real-time learning for data-driven dynamical systems. Specifically, we have identified the following data-rich dynamical engineering systems: power and energy systems, transportation systems, as well as signal and information processing systems. Topics for discussions include various real time learning approaches for the engineering systems (such as deep learning architectures, model-based learning, model-free learning, reinforcement learning, etc.), data representation for engineering systems (including the research problems in feature extraction, graphical models, real time unsupervised learning, etc.), and the potential solutions for closing the loop around data. The workshop will examine control, signal processing, machine learning, communication, power and energy, and transportation systems. Intellectual Merit: The workshop will address key questions in real-time learning and decision making that arises from dynamic data-driven engineering systems. Topics of interest include different perspectives of learning and decision making from control, signal processing, machine learning, computational intelligence, and domain applications' viewpoint. They will synergistically contribute towards two of the ten big ideas from NSF; ``Harnessing Data for 21st Century Science and Engineering'' and "Growing Convergent Research at NSF''. Broader Impacts: This workshop will engage and promote convergent research centered around real-time decision making and data in engineering systems. The discussions will have significant impact on the research and education of machine learning and large-scale data-driven engineering systems. The workshop will also generate opportunities for collaborative research between academia and industry. Ideas will also be generated for possible innovation and entrepreneurship. The workshop report will serve as an important resource for the scientific community and the general public.