Who
Dr. Anastasios Tsiamis
When
21 July 2025, 11:00 Athens time, Science Building 145Π42
Title
Learning Theory for Control: Algorithms, Rates, Fundamental Limits
Abstract
Machine learning is poised to play an increasingly central role in the future of autonomous systems. However, deploying learning-based methods safely and reliably in the real world requires a principled and integrated theoretical understanding of learning-based control. In this talk, I will present recent progress toward this goal, drawing on tools from both systems theory and learning theory. In the first part of the talk, we will explore the fundamental limits of learning-based control: what makes a system easy or hard to learn? Our focus will be on sample complexity - the minimum number of samples required to accurately learn a model or control policy. We will show how system-theoretic properties such as controllability can significantly influence the learning process. In particular, we will demonstrate that systems with poor controllability structure - such as underactuated systems - can exhibit provably high sample complexity, regardless of the learning algorithm used. Time permitting, the second part of the talk will discuss the use of online learning techniques for adaptive control in dynamic environments. We will focus on the problem of online tracking control of unknown and moving targets, which may be non-stationary and revealed only sequentially. By leveraging online learning methods, we can design control algorithms that come with theoretical performance guarantees despite the nonstationarity. We will demonstrate the practical effectiveness of these methods through experiments on a real quadrotor platform.
About the Speaker
Anastasios Tsiamis received the Diploma degree in electrical and computer engineering from the National Technical University of Athens, Greece, in 2014. He obtained his PhD at the Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA, USA, in 2022. Currently, he is a senior scientist at the Automatic Control Laboratory, ETH Zurich, Switzerland. His research interests include statistical and online learning in the setting of control systems, as well as robust and risk-aware control. Anastasios Tsiamis was a finalist for the IFAC Young Author Prize in IFAC 2017 World Congress and a finalist for the Best Student Paper Award in ACC 2019. He is a coauthor to the paper that has won the Best Student Paper Award in CDC 2022.
