A Hybrid Control Framework for Accelerated Methods with Exponential Rate
Event details
| Date | 25.05.2018 |
| Hour | 10:15 › 11:00 |
| Speaker | Tamás Keviczky, Delft Center for Systems and Control, Delft University of Technology |
| Location | |
| Category | Conferences - Seminars |
Abstract: Ordinary differential equations, and in general a dynamical system viewpoint, have seen a resurgence of interest in developing fast optimization methods, mainly thanks to the availability of well-established analysis tools.
In this talk, I will provide an overview of fast algorithms and recent results from a dynamical systems perspective. I will then describe a hybrid control framework to design a class of fast gradient-based methods in continuous-time that, in comparison with the existing literature including Nesterov’s fast-gradient method, features a state-dependent, time-invariant damping term that acts as a feedback control input. The proposed design scheme allows for a user-defined, exponential rate of convergence for a class of nonconvex, unconstrained optimization problems. Finally, I will introduce a discretization method such that the resulting discrete dynamical system possesses an exponential rate of convergence.
Bio: Dr. Tamas Keviczky is an Associate Professor in Networked Cyber-Physical Systems at the Delft Center for Systems and Control, TU Delft. He was a Postdoctoral Scholar at the California Institute of Technology and received his PhD in Control Science and Dynamical Systems from the University of Minnesota. He was awarded the AACC O. Hugo Schuck Best Paper Award for Practice. He has served as an Associate Editor of Automatica since 2011 and has published over 100 scientific articles. His main research interests include distributed optimization and optimal control, model predictive control, embedded optimization-based control and estimation of large-scale systems with applications in aerospace, automotive and mobile robotics, industrial processes, and infrastructure systems such as water, heat, and electricity networks.
In this talk, I will provide an overview of fast algorithms and recent results from a dynamical systems perspective. I will then describe a hybrid control framework to design a class of fast gradient-based methods in continuous-time that, in comparison with the existing literature including Nesterov’s fast-gradient method, features a state-dependent, time-invariant damping term that acts as a feedback control input. The proposed design scheme allows for a user-defined, exponential rate of convergence for a class of nonconvex, unconstrained optimization problems. Finally, I will introduce a discretization method such that the resulting discrete dynamical system possesses an exponential rate of convergence.
Bio: Dr. Tamas Keviczky is an Associate Professor in Networked Cyber-Physical Systems at the Delft Center for Systems and Control, TU Delft. He was a Postdoctoral Scholar at the California Institute of Technology and received his PhD in Control Science and Dynamical Systems from the University of Minnesota. He was awarded the AACC O. Hugo Schuck Best Paper Award for Practice. He has served as an Associate Editor of Automatica since 2011 and has published over 100 scientific articles. His main research interests include distributed optimization and optimal control, model predictive control, embedded optimization-based control and estimation of large-scale systems with applications in aerospace, automotive and mobile robotics, industrial processes, and infrastructure systems such as water, heat, and electricity networks.
Practical information
- General public
- Free