BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:DLAB Seminar: Ordinal latent variables for learning intensity scal
 es\, by Niklas Stoehr
DTSTART:20220530T144500
DTEND:20220530T153000
DTSTAMP:20260411T182817Z
UID:5e165c451f0bfcfd0a284f07ddd6278d53a33dcc9a005d760f75f057
CATEGORIES:Conferences - Seminars
DESCRIPTION:Niklas Stoehr \nConcepts such as conflict intensity and sentim
 ent polarity describe an intrinsic ordering\, but are subjective and abstr
 act. Thus\, they may be best described using ordinal scales. They are howe
 ver frequently treated as either nominal (categorical) or cardinal (contin
 uous). We propose modelling intensity concepts using an ordinal latent var
 iable model that can be learned from observable correlates. Particularly\,
  we focus on conflict intensity: we learn an event-level intensity scale f
 rom observed perpetrator\, victim and action types as well as fatality cou
 nts. Going beyond the level of individual events\, we incorporate temporal
  dimensions of perceived intensity drawing connections to media attention 
 and the notion of surprise.\n\nNiklas Stoehr is a doctoral student at the 
 Institute for Machine Learning at ETH Zurich advised by Ryan Cotterell and
  Bob West at EPFL. His interdisciplinary research aims at measuring laten
 t intensity concepts. Combining methods from Natural Language Processing
 \, Computational Social Science and Network Science\, his interests are p
 articularly centred around conflict intensity\, numbers in text and sentim
 ent analysis. He previously worked at the interface of these fields in ind
 ustry (IBM AI Core\, Microsoft Research\, German Federal Foreign Office) a
 nd research (University College London\, University of Oxford\, Tsinghua U
 niversity\, TU Berlin).
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR