BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:Europe/Vienna
BEGIN:DAYLIGHT
DTSTART:20190331T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20181028T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260405T191640Z
UID:5c8a5b8a62e90908133539@ist.ac.at
DTSTART:20190327T130000
DTEND:20190327T141500
DESCRIPTION:Speaker: Chris Hofer\nhosted by Herbert Edelsbrunner\nAbstract:
  We present some recent applications of persistent homologyin machine lear
 ning. First\, we introduce a metric shape spacebased on a topological repr
 esentation of 2D/3D objects.The metric allows to use classic metric-based 
 machine learningalgorithms\, e.g.\, k-nearest neighbors or k-means cluster
 ing.Next\, we establish a relation between end-to-end learnabledeep neural
  networks and persistence barcodes.The key contribution here is the constr
 uction of parametrizedvectorization schemes which respect the stability pr
 operties ofpersistent homology computation. These vectorization schemescan
  be implemented as a learnable input layer for neuralnetworks\, yielding a
 n approach for supervised end-to-end learningin the regime of persistence 
 barcodes.Finally\, we leverage that Vietoris-Rips persistent homology is l
 ocally differentiable and apply this insight to impose topological constra
 ints on the latent representations learned by an autoencoder.These represe
 ntations show beneficial propertiesfor kernel density based one-class lear
 ning.
LOCATION:Mondi Seminar Room 3\, Central Building\, ISTA
ORGANIZER:hwagner@ist.ac.at
SUMMARY:Chris Hofer: GeomTop Seminar: Applying Persistent Homology in Machi
 ne Learning
URL:https://talks-calendar.ista.ac.at/events/1880
END:VEVENT
END:VCALENDAR