BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:AI+Pizza July 2018 - John Bronskill - University of Cambridge\, Ri
 anne Van den Berg\, Microsoft Research
DTSTART:20180726T163000Z
DTEND:20180726T180000Z
UID:TALK108508@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:*Speaker 1* - John Bronskill\n*Title* - VERSA: A Versatile and
  Efficient Framework for Few-shot Learning\n*Abstract* - VERSA is a new\, 
 probabilistic\, amortized\, multi-task\, meta-learning framework. It achie
 ves state-of-the-art performance in one-shot learning on the Omniglot and 
 miniImageNet datasets\, and produces compelling results on a one-shot Shap
 eNet view reconstruction challenge. This talk will give an overview of the
  system\, describe how it works\, and present results on few-shot classifi
 cation and view reconstruction tasks.\n*Speaker 2* - Rianne Van den Berg\n
 *Title* - Sylvester Normalizing Flows for Variational Inference\n*Abstract
 * - Variational inference relies on flexible approximate posterior distrib
 utions. Normalizing flows provide a general recipe to construct flexible v
 ariational posteriors. We introduce Sylvester normalizing flows\, which ca
 n be seen as a generalization of planar flows. Sylvester normalizing flows
  remove the well-known single-unit bottleneck from planar flows\, making a
  single transformation much more flexible. We compare the performance of S
 ylvester normalizing flows against planar flows and inverse autoregressive
  flows and demonstrate that they compare favorably on several datasets.\n\
 nJoint work with Leonard Hasenclever\, Jakub M Tomczak\, Max Welling\, to 
 be presented at UAI 2018.\nhttps://arxiv.org/abs/1803.05649\n\n
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR