BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Omer Eryilmaz (University of Birmingham)
DTSTART:20260316T113000Z
DTEND:20260316T133000Z
DTSTAMP:20260423T024549Z
UID:BNAT/18
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/BNAT/18/">Fl
 ow-Aware Ellipsoidal Filtration for Persistent Homology of Recurrent Signa
 ls</a>\nby Omer Eryilmaz (University of Birmingham) as part of Basic Notio
 ns and Applied Topology Seminar\n\nLecture held in Room 1 at the IMPAS\, R
 oom 1.14 at the Institute of Informatics (University of Gdańsk).\n\nAbstr
 act\nRecurrent signals give rise to trajectories that repeatedly return cl
 ose to earlier states in state space. Analysing such data therefore requir
 es a principled notion of similarity between states. In practice\, this de
 pends on how local neighbourhoods are defined and scaled. These neighbourh
 oods are also important for topology-preserving denoising in state space\,
  where the aim is to reduce noise without distorting the underlying trajec
 tory structure. This talk introduces a flow-aware ellipsoidal filtration f
 or persistent homology based on a spatio-temporal covariance construction 
 that estimates local flow geometry from both temporal and spatial neighbou
 rs. Unlike isotropic constructions based on balls\, such as the Vietoris--
 Rips filtration\, the proposed method assigns an ellipsoid to each point\,
  with orientation and axis lengths determined by local flow variances. Whe
 n a dominant $H_1$ feature captures the main recurrent loop structure\, it
 s persistence interval can be used as a data-driven scale selection rule. 
 Experiments on synthetic and real signals suggest that flow-aware ellipsoi
 dal neighbourhoods can improve topology-preserving denoising and first-rec
 urrence-time estimation compared with the Vietoris--Rips filtration. More 
 broadly\, the results illustrate how incorporating anisotropy into persist
 ent homology can provide a more informative description of recurrent dynam
 ical systems.\n
LOCATION:https://researchseminars.org/talk/BNAT/18/
END:VEVENT
END:VCALENDAR
