Learning and decision-making with data: Optimal formulations and phase transitions

A Bennouna, BPG Van Parys - arXiv preprint arXiv:2109.06911, 2021 - arxiv.org
We study the problem of designing optimal learning and decision-making formulations when
only historical data is available. Prior work typically commits to a particular class of data …

Efficient learning of a linear dynamical system with stability guarantees

W Jongeneel, T Sutter, D Kuhn - IEEE Transactions on …, 2022 - ieeexplore.ieee.org
We propose a principled method for projecting an arbitrary square matrix to the nonconvex
set of asymptotically stable matrices. Leveraging ideas from large deviations theory, we …

Data-Driven and Stealthy Deactivation of Safety Filters

D Arnström, AMH Teixeira - arXiv preprint arXiv:2412.01346, 2024 - arxiv.org
Safety filters ensure that control actions that are executed are always safe, no matter the
controller in question. Previous work has proposed a simple and stealthy false-data injection …

On topological equivalence in linear quadratic optimal control

W Jongeneel, D Kuhn - 2021 European Control Conference …, 2021 - ieeexplore.ieee.org
Dynamical systems are topologically equivalent when their orbits can be mapped onto each
other via a homeomorphic change of coordinates. We will show that in general, closed-loop …

Efficient Robustness and Interpretability in Learning and Data-Driven Decision-Making

MA Bennouna - 2024 - dspace.mit.edu
As machine learning algorithms are increasingly developed and deployed in high-stakes
applications, ensuring their reliability has become crucial. This thesis introduces algorithmic …

Topological Obstructions

W Jongeneel, E Moulay - … Obstructions to Stability and Stabilization: History …, 2023 - Springer
Given the previous chapters on topology and dynamical system theory, we can now provide
a consistent treatment of topological obstructions to stability and stabilization. In particular …