唐漾，博士，教授，博士生导师，德国洪堡基金、国家级高层次人才、科技部中青年科技创新领军人才、上海市优秀学术带头人等计划入选者,ESI全球高被引科学家。主要研究多智能体系统/复杂网络状态估计、控制和优化，自主群体智能系统感知和决策，机器视觉与深度学习，信息物理融合系统安全分析与控制，以及过程系统风险预警和应急辅助决策。围绕上述领域，在Nature子刊、Cell子刊、Automatica和IEEE汇刊上发表论文100余篇，申请/公开/授权专利10件。目前担任Nature出版集团Scientific Reports资深编委，IEEE Transactions on Neural Networks and Learning Systems, IEEE Transactions on Cybernetics, IEEE Transactions on Circuits and Systems I: Regular Papers, IEEE Transactions on Emerging Topics in Computational Intelligence, IEEE Systems Journal和Engineering Applications of Artificial Intelligence (IFAC Journal)等多个SCI期刊的副主编/编委。获得2019年度上海市自然科学奖一等奖（第一完成人）。

报告摘要：多智能体系统可以通过相互协作和优化来解决单个智能体难以或无法解决的复杂系统科学问题。在多智能体系统中，具有一定自主性的智能体由于相关性与协调性而产生复杂的交互作用，表现为合作/竞争行为。本报告主要关注多智能体合作优化与合作/非合作博弈。首先从合作优化入手，总结了分布式优化和联邦优化的相关研究，侧重于介绍分布式在线优化与联邦优化及其在隐私保护中的应用；接着，引入合作博弈和非合作博弈，分别关注群体合作行为的涌现以及个体行为与收益，扩展合作优化的相关问题；然后，介绍了优化与博弈相关的实际应用；最后，讨论了合作优化、合作/非合作博弈的未来方向。新的研究进展揭示了借助优化与博弈方法有助于对多智能体系统的内在联系获得更深刻的理解，可期望为复杂系统科学问题提供更有效的分析工具。

李国齐，博士生导师，2011年毕业于新加坡南洋理工大学,获博士学位。2011年-2014年就职于新加坡科技局，任科学家; 2014年-2022年任教于清华大学, 历任助理研究员、助理教授、副教授、博士生导师; 2022年3月入职中国科学院自动化所。李国齐在Nature、Nature Communications、Proceedings of the IEEE、IEEE TPAMI等期刊和ICLR、NeurIPS、AAAI、CVPR等会议上发表论文 150余篇；出版国内类脑计算领域早期学术专著1部；论文在 Google 学术上被引用 4600余次；申请专利 33 项（转让6项）；主持自然基金委基金项目、科技部重点研发项目、JKW基础加强项目、北京市科委项目等20余项；担任多个国际会议的 Tutorial Chair，Symposium Chair，Track Chair，Publicity Chair 和组委会成员；担任多个国内期刊和国际SCI期刊的编委、副主编；2017 年入选北京市自然科学基金优秀青年人才，2018 年获得中国指挥与控制学会科学技术一等奖，2019 年入选北京智源人工智能研究院“智源学者”，2021年获得福建省科技进步二等奖，2021 年获得北京市杰出青年基金资助; 其参与的类脑芯片理论、架构和工具链的工作曾入选2019年中国科学十大进展和2020年世界人工智能十大进展。

报告摘要：2011年一篇发表于Nature工作提出利用图论的最大匹配（Maximum Matching）算法寻找最小的控制源个数以保证网络结构可控性的方法，并给出了所需独立控制源个数的下界。本报告的工作指出可以通过时空切换输入的策略，进一步将最小的独立控制源数目减少到2个。我们基于一个类似于人类神经系统（秀丽隐杆线虫）的复杂网络，仅使用最大匹配算法给出的控制源个数下界6%数目的控制源，就可保证整个网络的结构可控性。

左宗玉教授，北京航空航天大学自动化科学与电气工程学院教授，博士生导师，教育部青年长江学者。主要从事非线性控制、无人飞行器控制理论及应用的相关研究工作，已出版英文专著2部、授权国家发明专利3项、发表SCI期刊论文80余篇，曾获天津市自然科学二等奖、国防技术发明三等奖以及军队科技进步三等奖各一项，曾入选科睿唯安2020年“全球高被引学者”以及爱思唯尔2021年“中国高被引学者”；目前担任北京航空航天大学自动化学院“理论力学A”的课程负责人，同时承担国家精品课、国家级一流课程“自动控制原理A”的教学工作；目前担任Journal of The Franklin Institute、Journal of Vibration and Control、International Journal of Aeronautical and Space Sciences等国际期刊以及中文核心期刊《系统科学与数学》编委，以及IEEE/CAA Journal of Automatica Sinica第一届青年编委。

报告摘要：With the rapid development of computing, automatic control and communication technologies, research and development on unmanned aerial vehicles (UAVs) have attracted extensive attention from all over the world during the last decades. Particularly due to the increasing demand of various civil applications, the conceptual design of UAV and the enabling flight control technology have been promoted and developed mutually. This talk is devoted to providing a brief introduction of some key elements of UAV control design, including motion equations, various classical and advanced control approaches, as well as the basic ideas, applicable conditions, advantages and disadvantages of these control approaches. Finally, some challenges and future research directions are raised.

Xinlei Yi received Ph.D. degree in electrical engineering from the School of Electrical Engineering and Computer Science, KTH Royal Institute of Technology in 2020 and then was a postdoc at the same university. He received B.S. and M.S. degrees in mathematics from China University of Geoscience and Fudan University, in 2011 and 2014, respectively. He receives a Wallenberg Foundation and WASP Postdoctoral Scholarship at MIT and will joint MIT in Fall 2022.

His current research interests include online optimization, distributed optimization, and event-triggered control. He is currently a member of early career advisory board of IEEE/CAA Automatica Sinica, and a member of IEEE Control Systems Society Technical Committee on Networks and Communication Systems and IFAC Technical Committee on Stochastic Systems. He received Best Paper Award in the 2021 International Conference on Industrial Artificial Intelligence and is a finalist for Zhang Si-Ying (CCDC) Outstanding Youth Paper Award in 2016.

报告摘要：The distributed nonconvex optimization problem of minimizing a global cost function formed by a sum of n local cost functions by using local information exchange is considered. This problem is an important component of many machine learning techniques with data parallelism, such as deep learning and federated learning. We propose a distributed primal–dual stochastic gradient descent (SGD) algorithm, suitable for arbitrarily connected communication networks and any smooth (possibly nonconvex) cost functions. We show that the proposed algorithm achieves the linear speedup convergence rate O(1/√nT) for general nonconvex cost functions and the linear speedup convergence rate O(1/(nT)) when the global cost function satisfies the Polyak–Łojasiewicz condition, where T is the total number of iterations. We also show that the output of the proposed algorithm with constant parameters linearly converges to a neighborhood of a global optimum. We demonstrate through numerical experiments the efficiency of our algorithm in comparison with the baseline centralized SGD and recently proposed distributed SGD algorithms.