BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.4.0.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:9073@i2m.univ-amu.fr
DTSTART;TZID=Europe/Paris:20260629T140000
DTEND;TZID=Europe/Paris:20260629T153000
DTSTAMP:20260622T070500Z
URL:https://www.i2m.univ-amu.fr/evenements/seminaire-de-pierre-latouche/
SUMMARY:Pierre Latouche (UCA): Séminaire de Pierre Latouche
DESCRIPTION:Pierre Latouche: Title: Importance weighted variational graph a
 utoencoders for inference in deep latent position block models\n\nAbstract
 :\n\nIn this presentation\, I will first show how latent position models c
 an be made compatible with block models for network analysis through the u
 se of deep generative models. I will focus on the binary edge case and int
 roduce a new random graph model along with a variational graph auto encodi
 ng strategy for inference. I will discuss the identifiability of the model
  and explain how model selection can be performed. I will then move to the
  sparse discrete edge case in the same deep\, block compatible\, modelling
  framework. I will show how importance weighted variational inference can 
 strongly improve the inference procedure over the classical variational au
 to encoding strategy. I will give the importance weights used for sampling
  and show that\, in the limit\, the lower bound obtained converges to the 
 integrated log likelihood of the data. Through this presentation\, I will 
 emphasise the need to rely on flexible random graph models to obtain relev
 ant loss functions.
CATEGORIES:Séminaire,Statistique
LOCATION:Saint-Charles - FRUMAM  (2ème étage)\, 3 Place Victor Hugo\, Mar
 seille\, 13003\, France
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=3 Place Victor Hugo\, Marse
 ille\, 13003\, France;X-APPLE-RADIUS=100;X-TITLE=Saint-Charles - FRUMAM  (
 2ème étage):geo:0,0
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:DAYLIGHT
DTSTART:20260329T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR