Наредни састанак Семинара биће одржан онлајн у среду, 3. маја 2023, са почетком у 19 часова.
Предавач: Владимир Костић, Faculty of Science, University of Novi Sad; Istituto Italiano di Tecnologia, Genova
Наслов предавања: AI FOR SCIENCE: STATISTICAL LEARNING PERSPECTIVE TO KOOPMAN OPERATOR THEORY FOR DATA-DRIVEN DYNAMICAL SYSTEMS
Апстракт: We are witnessing striking progress originating from the use of AI, and, in particular, Machine Learning (ML) technologies in science. Traditional scientific modelling by equations of motion is being more and more powered by complementary data-driven approaches to solve challenging problems such as protein folding and fluid-dynamics. In this quickly growing field, the theory of Koopman operators have found particularly prominent place due to the fact that non-linear dynamical systems can be handily described by the associated linear (Koopman) operators whose action evolves observables of a system forward in time. While data-driven algorithms to reconstruct such operators are now well known, their relationship with statistical learning is still largely unexplored. To bridge this gap, in this talk we will present a framework to learn Koopman operators from finite data trajectories using reproducing kernel Hilbert spaces (RKHS). Doing this, we provide high-probability finite-sample theoretical learning guarantees, which are of the paramount importance for safe and trustworthy employment of the AI based on the Koopman operator theory in scientific applications.
Напомена: Регистрациона форма за учешће на Семинару је доступна на линку:
https://miteam.mi.sanu.ac.rs/asset/CW5nJWDSEZDj7p32p
Уколико желите само да пратите предавање без могућности активног учешћа, пренос је доступан на линку:
https://miteam.mi.sanu.ac.rs/asset/4LNW8WtML7rLKojoz