BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260624T014717EDT-5502NzjT5R@132.216.98.100
DTSTAMP:20260624T054717Z
DESCRIPTION:THIS SEMINAR HAS BEEN CANCELLED\n\nTitle: New Perspectives on G
 enerative Adversarial Networks\n\nAbstract: Generative Adversarial Network
 s (GANs) are a popular generative modeling approach known for producing ap
 pealing samples\, but their theoretical properties are not yet fully under
 stood\, and they are notably difficult to train. In the first part of this
  talk\, I will provide some insights on why GANs are a more meaningful fra
 mework to model high dimensional data like images than the more traditiona
 l maximum likelihood approach\, interpreting them as “parametric adversari
 al divergences” and rooting the analysis with statistical decision theory.
  In the second part of the talk\, I will address the difficulty of trainin
 g GANs from the optimization perspective by importing tools from the mathe
 matical programming literature. I will survey the “variational inequality”
  framework which contains most formulations of GANs introduced so far\, an
 d present theoretical and empirical results on adapting the standard metho
 ds (such as the extragradient method) from this literature to the training
  of GANs.\n	The talk is based on the following two papers: “Parametric Adve
 rsarial Divergences are Good Task Losses for Generative Modeling”\, G. Hua
 ng\, H. Berard\, A. Touati\, G. Gidel\, P. Vincent\, S. Lacoste-Julien htt
 ps://arxiv.org/abs/1708.02511\n	“A Variational Inequality Perspective on GA
 Ns”\, G. Gidel\, H. Berard\, G. Vignoud\, P. Vincent\, S. Lacoste-Julien h
 ttps://arxiv.org/abs/1802.10551 to appear at ICLR 2019\n	 \n
DTSTART:20190125T203000Z
DTEND:20190125T213000Z
SUMMARY:CANCELLED--------Simon Lacoste-Julien\, Université de Montréal
URL:/mathstat/channels/event/cancelled-simon-lacoste-j
 ulien-universite-de-montreal-293447
END:VEVENT
END:VCALENDAR