BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Learning Without Labels: Manifold Geometry for Medical Image Segme
 ntation
DTSTART:20260430T060000Z
DTEND:20260430T080000Z
DTSTAMP:20260504T021800Z
UID:indico-event-9334@scitalks.tifr.res.in
DESCRIPTION:Speakers: Tripti Bameta (ACTREC\, Tata Memorial Center)\n\nMed
 ical image analysis is constrained less by algorithms than by the cost of 
 expert annotation. I will describe an alternative paradigm in which segmen
 tation emerges from the geometry of pretrained neural representations rath
 er than from labelled training. The pipeline  embed\, project with UMAP\,
  cluster via the graph Laplacian  is simple\, modular\, and requires only
  light post-hoc labelling. I will focus on HistoPAINT\, our histopathology
  application\, and close with thoughts on where geometric\, low-supervisio
 n pipelines are likely to displace fully supervised ones in scientific ima
 ging.\n\nhttps://scitalks.tifr.res.in/event/9334/
LOCATION:AG 66 and on Zoom
URL:https://scitalks.tifr.res.in/event/9334/
END:VEVENT
END:VCALENDAR