Talks

Schedule

9:00 | Opening

9:05 | Talk 1: Geethu Joseph

9:30 | Talk 2: Jing Shuang (Lisa) Li

10:00 | Talk 3: Victor Magron

10:30 | Coffee break

11:00 | Talk 4: René Vidal

11:30 | Talk 5: Masaaki Nagahara

12:00 | Panel discussion: Challenges, Opportunities, Outlook

12:20 | Closing Remarks

12:30 | End of workshop

Speakers

Geethu Joseph

Title: Sparse Actuator Control: Controllability and Design Algorithms

Constraining the control inputs to be sparse is often necessary to select a small subset of available actuators at each time instant due to energy, bandwidth, or physical network constraints. This talk explores how sparsity constraints on actuator use impact the controllability of linear dynamical systems. We derive the necessary and sufficient conditions for controllability under sparsity constraints and examine how classical control theory and compressed sensing converge to optimize control strategies in large-scale networks. Ultimately, this talk presents a theoretical basis for sparsity-constrained systems and establishes connections to other algorithmic approaches.

Short bio

Geethu Joseph received the B. Tech. degree in electronics and communication engineering from the National Institute of Technology, Calicut, India, in 2011, and the M. E. degree in signal processing and the Ph.D. degree in electrical communication engineering (ECE) from the Indian Institute of Science (IISc), Bangalore, in 2014 and 2019, respectively. She was a postdoctoral fellow with the Department of Electrical Engineering and Computer Science at Syracuse University, NY, USA, from 2019 to 2021. She is currently a tenured assistant professor in the signal processing systems group at the Delft University of Technology, Delft, Netherlands.

Dr. Joseph was awarded the 2022 IEEE SPS Best PhD dissertation award and the 2020 SPCOM Best Doctoral Dissertation award. She is also a recipient of the Prof. I. S. N. Murthy Medal in 2014 for being the best M. E. (signal processing) student in the ECE dept., IISc, and the Seshagiri Kaikini Medal for the best Ph.D. thesis of the ECE dept. at IISc for the year 2019-'20. She is an associate editor of the IEEE Sensors Journal. Her research interests include statistical signal processing, network control, and machine learning.

Jing Shuang (Lisa) Li

Title: Global Performance Guarantees for MPC Under Sparse Local Communication

Recent advances in model predictive control (MPC) leverage sparse, local communication (as opposed to global, or centralized communication) to produce scalable distributed MPC algorithms. In this talk, I will present theories and simulations that explore how sparse communication affects performance. A striking finding is that in a network of 121 coupled pendula if we restrict each node to only communicate with its immediate neighbors, the resulting control can still be globally optimal --- thus, the inclusion of severe communication constraints need not compromise performance. I will also briefly discuss the connections between sparse control and network neuroscience.

Short bio

Jing Shuang (Lisa) Li received a Bachelor of Applied Science from the University of Toronto and a PhD in Control and Dynamical Systems from Caltech. She joined the University of Michigan as an Assistant Professor in the fall of 2023. Her research interests include distributed control, resource- and communication-constrained control, and large-scale systems with applications in engineering and neuroscience.

Victor Magron

Title: Sparse Polynomial Optimization for Instability Certification of ReLU Feedback Systems

Polynomial optimization methods often encompass many major scalability issues on the practical side. Fortunately, for many concrete problems, we can look at them in the eyes and exploit the inherent data structure arising from the input cost and constraints. We consider the stability analysis of feedback systems with rectified linear unit (ReLU) activations and model this problem with polynomial optimization. We construct a complete hierarchy of convex programs allowing one to certify instability. Then, we leverage the equality constraints arising from ReLU encoding to obtain a complete sparse hierarchy, yielding significant scalability improvements.

Short bio

Victor Magron is a junior researcher at the Laboratoire d'analyse et d'architecture des systèmes (LAAS-CNRS), in the department of Decision and Optimization, in Toulouse, France. He did a PhD in formal proofs for global optimization at Ecole Polytechnique, France, funded by INRIA, under the supervision of Benjamin Werner and Stéphane Gaubert. In 2014, he was a Postdoc in the Methods and Algorithms for Control (MAC) team from LAAS-CNRS under the supervision of Didier Henrion and Jean-Bernard Lasserre, funded by the Simone and Cino del Duca foundation of the Institut de France. During 2014-2015, he was a Research Associate in the Circuits and Systems group at Imperial College in collaboration with George A. Constantinides and Alastair Donaldson. From 2015 to 2018, he was a CNRS junior researcher affiliated with VERIMAG in Grenoble. In 2019, he was affiliated with LAAS-CNRS in the MAC team located in Toulouse. He has also been an associate researcher at the Institute of Mathematics from Toulouse since 2020. Since 2022 he has taken the lead of a new team in LAAS focusing on polynomial and moment optimization, called POP. His research is devoted to applications of certified optimization (e.g., semidefinite optimization), especially polynomial optimization to quantum and energy networks.

René Vidal

Title: Observability and Invertibility of Linear Systems with Sparse Inputs

In this talk, I will present recent results on the observability and invertibility of a new class of linear time-invariant systems, where the input at each time step is sparse with respect to an overcomplete dictionary. Our results show that the sparsity assumption enables the recovery of both the initial state and the sparse, yet unknown, input from output measurements alone. Specifically, we derive conditions for the existence and uniqueness of sparse inputs and provide necessary and sufficient conditions for a linear program to recover both the initial state and the sparse input. We also tackle the left inversion problem for systems with sparse inputs, introducing novel geometric characterizations that extend classical results to the sparse-input setting. Through numerical validation and a concrete example, we highlight the practical applications of these results in state estimation, efficient system inversion, and sparse control.

Short bio

René Vidal is the Penn Integrates Knowledge and Rachleff University Professor of Electrical and Systems Engineering \& Radiology and the Director of the Center for Innovation in Data Engineering and Science (IDEAS) at the University of Pennsylvania. He is also an Amazon Scholar, an Affiliated Chief Scientist at NORCE, and a former Associate Editor in Chief of TPAMI. His current research focuses on the foundations of deep learning and trustworthy AI and its applications in computer vision and biomedical data science. Dr. Vidal is an ACM Fellow, AIMBE Fellow, IEEE Fellow, IAPR Fellow, and Sloan Fellow, and has received numerous awards for his work, including the IEEE Edward J. McCluskey Technical Achievement Award, D’Alembert Faculty Award, J.K. Aggarwal Prize, ONR Young Investigator Award, NSF CAREER Award as well as best paper awards in machine learning, computer vision, signal processing, controls, and medical robotics.

Masaaki Nagahara

Title: Non-Convex Optimization Approach to Maximum Hands-off Control

This talk presents a novel computational method for sparse control, also known as maximum hands-off control, using non-convex penalty functions such as the minimax concave penalty. The sparse control is formulated as the L0-optimal control problem, which is difficult to directly solve. Conventionally, the L1-norm has been used as a surrogate for the L0 norm to numerically obtain the solution. However, the L1-norm approximation may not always yield sparse control. To overcome this difficulty, we propose non-convex functions such as the minimax concave penalty as a surrogate for the L0 norm. We establish the equivalence of the original and proposed control problems without relying on the normality assumption, which is typically required when approximating the L0 norm with the L1 norm. Furthermore, we present an effective numerical algorithm for the proposed optimal control based on the Alternating Direction Method of Multipliers. A design example is shown to illustrate the effectiveness of the proposed method.

Short bio

Masaaki Nagahara received a bachelor's degree in engineering from Kobe University in 1998 and a master's degree and a Doctoral degree in informatics from Kyoto University in 2000 and 2003, respectively. He is currently a Full Professor at the Graduate School of Advanced Science and Engineering, Hiroshima University. He has been a Visiting Professor at the Indian Institute of Technology Bombay since 2017. His research interests include control theory, machine learning, and sparse modeling. He received remarkable international awards: The Transition to Practice Award in 2012 and the George S. Axelby Outstanding Paper Award in 2018 from the IEEE Control Systems Society. Also, he received many awards from Japanese research societies, such as the SICE Young Authors Award in 1999, the SICE Best Paper Award in 2012, the SICE Best Book Authors Awards in 2016 and 2021, the SICE Control Division Research Award (Kimura Award) in 2020, and the Best Tutorial Paper Award from the IEICE Communications Society in 2014. He is a senior member of IEEE, and a member of IEICE, SICE, ISCIE, and RSJ.

Page updated

Google Sites

Report abuse