BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:The Mayer-Vietoris spectral sequence for topological data analysis
DTSTART:20261112T100000
DTEND:20261112T110000
DTSTAMP:20261109T140456Z
UID:4884afdfbdf619160866bd6f717108833af26e485c2f6fc018f11698
CATEGORIES:Conferences - Seminars
DESCRIPTION:Freya Jensen\, Max Planck Institute of Molecular Cell Biology 
 and Genetics\, Dresden\nPersistent homology is a central tool of topologic
 al data analysis (TDA)\, but its usual pipeline assumes an unstructured po
 int cloud. Often\, though\, the data come with extra structure: a cover! E
 xamples are huge simplicial complexes built in parallel across many comput
 e nodes\, or point clouds with labels (cell or molecule types) that group 
 the data. The Mayer-Vietoris spectral sequence (MVSS) is built for exactly
  this situation: it relates the persistent homology of the pieces of a cov
 er and their intersections to that of the whole. But can it actually be ma
 de into a useful tool for TDA?\nIn this talk I take you on that journey. I
  present a C++/MPI implementation of the MVSS for 3D complexes and show wh
 at it buys us\, notably a per-process memory footprint that shrinks as the
  cover grows. I then turn to a case study on labeled data\, where the spec
 tral sequence tells apart patterns without ever computing the full persist
 ent homology.
LOCATION:CM 1 517 https://plan.epfl.ch/?room==CM%201%20517 https://epfl.zo
 om.us/j/64555981838
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR