Learning for Dynamics and Control Workshop, To Be Held in Boston, MA, May 30-31, 2019

在过去的几年里，机器学习在计算机视觉和语言翻译等众多领域产生了巨大的影响。在接下来的十年里，最大的数据来源预计将是感知和控制物理世界的设备。这种从物理世界中涌现的实时数据的爆炸式增长，需要机器学习、基于模型的动态系统以及控制和决策理论等领域的融洽。虽然控制理论已经牢固地植根于基于模型的设计的传统，但数据的可用性和规模(包括时间和空间)将需要重新思考该学科的基础。从机器学习的角度来看，未来的主要挑战之一是超越模式识别，解决数据驱动的控制和决策以及基于学习的动态过程优化中的问题。我们建议于5月30日至31日在麻省理工学院举行为期两天的关于学习、动力学和控制之间的接口的成立会议。从长远来看，我们希望将这一活动作为年度会议，展示这一新兴界面的最佳成果。我们的目标是创建一个新的社区，人们严格地跨学科思考，提出新的基本问题，并为这一新的科学领域奠定基础。我们认为，我们提议的努力将导致一次非常有影响力的年度会议，可以吸引许多科学界的大量关注。这一提议的学术价值在于创建了一个科学论坛，将控制系统、优化、机器学习和相关学科领域的先驱和最先进的研究聚集在一起，创建了一个享有盛誉的年度会议，定义了动态和控制系统学习的最新水平。关于这一主题的精英会议不仅可以通过架起两个遥远地区的桥梁在科学上产生巨大影响，而且还可以从社区的角度产生巨大影响，培养越来越多致力于这一新兴界面的初级研究人员。一个涉及控制、优化和学习的精英会议将为学生、博士后和初级教师的专业发展提供一个自然和培养的家。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。

Over the past few years, machine learning has had tremendous impact in numerous areas such as computer vision and language translation. Over the next decade, the biggest generator of data is expected to be devices which sense and control the physical world. This explosion of real-time data that is emerging from the physical world requires a rapprochement of areas such as machine learning, model based dynamical systems, and control and decision theory. While control theory has been firmly rooted in tradition of model-based design, the availability and scale of data (both temporal and spatial) will require rethinking of the foundations for the discipline. From a machine learning perspective, one of the main challenges going forward is to go beyond pattern recognition and address problems in data driven control and decision making as well as learning-based optimization of dynamical processes.We propose a two-day inaugural conference on the interface between learning, dynamics and control to take place on May 30-31 at MIT. Long term we would like to sustain this event as an annual conference where best results on this emerging interface are presented. We aim to create a new community of people that think rigorously across the disciplines, ask novel fundamental questions, and develop the foundations of this new scientific area. We feel that our proposed effort will result in a very influential annual conference that can attract a lot of attention from numerous scientific communities. The intellectual merit of this proposal is the creation of a scientific forum that brings together pioneers and state of the art research in the areas of control systems, optimization, machine learning, and related disciplines to create a prestigious annual conference that defines the state-of-the-art in Learning for Dynamical and Control Systems. An elite conference on this topic can have tremendous impact not only scientifically by bridging two distant areas but also from a community perspective that nurtures a growing number of junior researchers working on this emerging interface. An elite conference across control, optimization and learning will provide a natural and nurturing home for the professional development for students, postdocs, and junior faculty.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.