Associated Researcher
 |
 |
 |
 |
 |
|
| I got my PhD on May 14th, 2021 from UAM under Marie Skłodowska-Curie fellowship at Conflex project. My research interests are focused on the control-theoretical aspects of free boundary problems arising in fluid mechanics, and more recently, deep learning from a mathematical control perspective. | |||||
Borjan Geshkovski is an Associated Researcher from UAM – Universidad Autónoma de Madrid under a Marie Skłodowska-Curie fellowship at Conflex Project. He earned a MSc in Mathematics at the University of Bordeaux, during which he did an internship on the topic “Obstacle problems: theory and Applications” within the DyCon team. He got his PhD in Control Theory under the supervision of Prof. Enrique Zuazua (FAU, University of Deusto and Universidad Autónoma de Madrid).
PhD thesis -slides (May 14th, 2021)
- MSc in Applied Mathematics (2016 – 2018). Université de Bordeaux, France
- BSc in Applied Mathematics and Computer Science (2012 – 2016). Université de Bordeaux, France
Master’s Thesis. Obstacle Problems: Theory and Applications
Advisor: Prof. Enrique Zuazua
Abstract: In this master thesis, we present a study of the analytical and optimal control properties for the elliptic and parabolic obstacle problems. The obstacle problem is one of the simplest and most physically relevant free boundary problems. From a mathematical perspective, similar questions as for classical partial differential equations (well-posedness, regularity, optimal control) are addressed, as well as the conception of appropriate numerical schemes and computer simulations of the these problems. This work was supported by the Advanced Grant DyCon (Dynamic Control) of the European Research Council Executive Agency (ERC).
Released
Turnpike in Optimal Control PDES, ResNets, and beyond
Turnpike in Lipschitz-nonlinear optimal control
Controllability of one-dimensional viscous free boundary flows
Null-controllability of perturbed porous medium gas flow
Accepted
Submitted
Sparse approximation in learning via neural ODEs
Large-time asymptotics in deep learning
Control and Deep Learning: Some connections
Awarded the “Best Review and Presentation Prize” at the second ConFlex workshop held in Bilbao, from February 11 to 21, 2019
- 09.12.2020 The interplay of Deep Learning and Control Theory, AG Mathematics of Deep Learning, FAU Erlangen-Nürnberg (Germany) | PDF Slides
- 23.10.2020 Large-time asymptotics in Deep Learning, Seminario de Estadísticas. UAM | PDF Slides
- 24.08.2020 Turnpike Control and Deep Learning, 2nd. Symposium on Machine Learning and Dynamical Systems. Fields Institute | PDF Slides
- 30.06.2020 Control in Interfaces and Deep Learning, 3rd. ConFlex workshop. Imperial College London | PDF Slides
- 21.01.2020 Mathematical Control and Deep Learning, FAU – Friedrich Alexander Universität, Erlangen-Nürnberg (Germany) | PDF Slides
- 23.08.2019 Control of perturbed porous medium flow, 8th Workshop on PDE, Optimal Design and Numerics, Centro de Ciencias “Pedro Pascual”, Benasque, Spain | PDF Slides
- 06.05.2019 Control of linearized porous medium flow, Workshop on homogenization, spectral theory and other topics in PDEs, ICMAT Madrid, Spain. PDF Slides
- 20.02.2019 Control of free boundary problems, Second Network meeting of the ConFlex consortium, Bilbao, Spain. PDF Slides
- 20.01.2019 Control and free boundaries, Friedrich-Alexander Universität, Erlangen, Germany. PDF Slides
- 27.04.2018 Obstacle problems, optimal control and numerics, DeustoTech, Bilbao, Spain. PDF Slides
Meet our team!
