BARC talk by Jacobus Conradi

Monday, March 31, 2025, Jacobus Conradi, PhD student at the University of Bonn, Germany, will give a talk on "On Research on Curve Similarities - Coresets, Partial matchings, and more".

Abstract:
Contemporary spatio-temporal data collection often produces vast datasets of (typically noisy) time-encoded data. In this talk, Jacobus Conradi will present results from his PhD thesis that address fundamental questions about handling large, potentially noisy datasets from a theoretical perspective. He will discuss topics ranging from clustering large sets of time-encoded data by first computing a coreset, despite the non-metric nature of the dynamic time warping distance, to identifying and clustering frequent patterns and their application to motion segmentation. Additionally, Jacobus will explore the computation of the Fréchet distance, even when the measured curves are interspersed with outliers."

Bio:
Jacobus is a fifth-year PhD student at the University of Bonn, supervised by Anne Driemel. His primary research focuses on similarity measures between curves and shapes. After completing my undergraduate studies in mathematics—where he primarily studied algebraic topology and graph theory—he joined the Theoretical Computer Science department in Bonn. Jacobus' current research lies in computational geometry, where he explores problems from both a theoretical and algorithm engineering perspective, such as clustering and nearest neighbor search when the objects in question are curves, as well as partial similarity between curves. Additionally, he has worked on problems motivated by practical applications, including polygon aggregation and non-obtuse triangulations.

Host:
Mikkel Vind Abrahamsen