Michael Reichel (ari)
Es spricht Prof. Dr. Martin Redmann (Universität Rostock)
zum Thema: Signature-Based Bilinear Approximations for Nonlinear Systems and Dimension Reduction
Abstract: Many real-world processes behave in complex, nonlinear ways and are difficult to analyze directly. In this talk, we present a method for approximating such systems by a related linear model that captures their behavior. Concepts from functional analysis are used to construct this linear approximation. The resulting model can be quite large, so we apply optimization techniques to reduce its size while preserving its essential dynamics. An advantage of the approach is that the model can be learned directly from data, without requiring explicit knowledge of the underlying equations. Numerical experiments illustrate the effectiveness of the method.
Dienstag, 19. Mai 2026, 15:00 Uhr, Curie Hörsaal
Alle Interessierten sind herzlich eingeladen!