ZJU-CSE Summer School 2021

Title(T/S II): Decentralized learning in the nonconvex world: Recent results
Speaker: Prof Hoi To Wai, Chinese University of Hong Kong
Date/Time: August 9, 2021, 10:00am-12:00pm, China Time (Check your local time here)
Abstract: Decentralized learning has become a critical enabler of the massively connected world that many people envision. In this talk, we discuss four key elements of scalable decentralized optimization and control: optimization problems, data, communication, and computation. We describe how these elements should work together in an effective and coherent manner. We review recent techniques developed for optimizing nonconvex models (i.e., problem classes) that process batch/streaming data (data types) across networks in a decentralized manner (communication and computation paradigm). We describe the intuitions and connections behind a core set of popular decentralized algorithms, emphasizing how to balance computation and communication costs. Practical issues and future research directions will also be discussed.
Speaker Homepage: https://www1.se.cuhk.edu.hk/~htwai/
Video Link(BiliBili), Slides

Title(T/S I):Distributed stochastic non-convex optimization: Optimal regimes and tradeoffs
Speaker: Prof Usman Khan, Tufts University
Date/Time: August 9, 2021, 8:30pm-10:00pm, China Time (Check your local time here)
Abstract: In many emerging applications, it is of paramount interest to learn hidden parameters from data. For example, self-driving cars may use onboard cameras to identify pedestrians, highway lanes, or traffic signs in various light and weather conditions. Problems such as these can be framed as classification, regression, or risk minimization in general, at the heart of which lies stochastic optimization and machine learning. In many practical scenarios, distributed and decentralized learning methods are preferable as they benefit from a divide-and-conquer approach towards data at the expense of local (short-range) communication. In this talk, I will present our recent work that develops a novel algorithmic framework to address various aspects of decentralized stochastic first-order optimization methods for non-convex problems. A major focus will be to characterize regimes where decentralized solutions outperform their centralized counterparts and lead to optimal convergence guarantees. Moreover, I will characterize certain desirable attributes of decentralized methods in the context of linear speedup and networkindependent convergence rates. Throughout the talk, I will demonstrate such key aspects of the proposed methods with the help of provable theoretical results and numerical experiments on real data.
Speaker Homepage: https://www.eecs.tufts.edu/~khan/
Video Link(BiliBili), Slides

Title(T/S III): Towards Scalable Algorithms for Distributed Optimization and Learning
Speaker: Prof. Cesar Uribe, Rice University
Date/Time: August 10, 2021, 8:30am-10:00am, China Time (Check your local time here)
Abstract: Increasing amounts of data generated by modern complex systems such as the energy grid, social media platforms, sensor networks, and cloud-based services call for attention to distributed data processing, in particular, for the design of scalable algorithms that take into account storage and communication constraints and help to make coordinated decisions. In this talk, we present recently proposed distributed algorithms with optimal convergence rates for optimization problems over networks, where data is stored distributedly. We focus on scalable algorithms and show they can achieve the same rates as their centralized counterparts, with an additional cost related to the structure of the network. We provide application examples to distributed inference and learning, and computational optimal transport.
Speaker Homepage: https://cauribe.rice.edu/
Video Link(BiliBili), Slides

Title(T/S IV): Distributed Optimization over Networks
Speaker: Prof Angelia Nedich, Arizona State University
Date/Time: August 10, 2021, 10:30am-12:00pm, China Time (Check your local time here)
Abstract: TBD






Speaker Homepage: https://angelia.engineering.asu.edu/

Title(T/S V): Bringing Statistical Thinking in Distributed Optimization. Vignettes from statistical inference over Networks (Part I, Part II)
Speaker: Prof Gesualdo Scutari, Purdue University
Date/Time: August 11, 2021, 8:30am-12:00pm, China Time (Check your local time here)
Abstract: There is growing interest in solving large-scale statistical machine learning problems over decentralized networks, where data are distributed across the nodes of the network and no centralized coordination is present (we termed these systems “meshed” networks). Modern massive datasets create a fundamental problem at the intersection of the computational and statistical sciences: how to provide guarantees on the quality of statistical inference given bounds on computational resources, such as time and communication efforts. While statistical-computation tradeoffs have been largely explored in the centralized setting, our understanding over meshed networks is limited: (i) distributed schemes, designed and performing well in the classical low-dimensional regime, can break down in the high-dimensional case; and (ii) existing convergence studies may fail to predict algorithmic behaviors; some are in fact confuted by experiments. This is mainly due to the fact that the majority of distributed algorithms over meshed networks have been designed and studied only from the optimization perspective, lacking the statistical dimension. Throughout some vignettes from low- and high-dimensional statistical inference, this talk goes over some designs and new analyses aiming at bringing statistical thinking in distributed optimization.
Speaker Homepage: https://engineering.purdue.edu/~gscutari/
Video Link(BiliBili)

Title(T/S VI): Distributed Load Frequency Control in Smart Grids
Speaker: Prof Shichao Liu, Carleton University
Date/Time: August 12, 2021, 8:30am-10:00am, China Time (Check your local time here)
Abstract: In a multi-area smart grid, the load frequency control is used to sustain the frequency of each control area and tie-line power at scheduled values through modifying power set points in case disturbances happen. For the cost-effective operation of the LFC, there are many challenges to be addressed such as controllers with limited energy resources and tremendous information exchanged among a large number of areas in the smart grid. In this talk, we firstly introduce smart grid control and communication structures. Then, the basics of load frequemcy control are covered briefly. At the end, we present distributed control and event-triggering schedule co-design schemes to tackle these challenges. These distributed control and scheduling coordination approaches can reduce the amount of needed information transmissions while not sacrificing the system dynamic performance.
Speaker Homepage: https://doe.carleton.ca/shichao-liu
Video Link(BiliBili), Slides

Title(T/S VII): Smart Sensing and Localization
Speaker: Prof Lihua Xie, Nanyang Technological University
Date/Time: August 12, 2021, 10:30am-12:00pm, China Time (Check your local time here)
Abstract: Sensing and localization are essential for IoT and intelligent unmanned systems. GPS has been widely used for positioning and navigation. However, in indoor and many outdoor environments such as urban canon, forest, tunnel, GPS may not be available or unreliable. Hence, there has been a lot of interest in developing technologies and algorithms for localization in such environments. In this talk, we shall discuss several sensing and localization systems and algorithms we have developed over the past few years including WiFi based indoor positioning and human activity recognition, UWB based localization, and vision-inertial-range sensor fusion for localization and mapping. Their applications in smart building/home, elderly care, UAV based structure inspection and AGV for logistics will be presented, and challenges and future research directions will be highlighted.
Speaker Homepage: https://www3.ntu.edu.sg/home/elhxie/index.html