Sylvester Normalizing Flows for Variational Inference

Open Access
Authors
Publication date 2018
Host editors
  • A. Globerson
  • R. Silva
Book title Uncertainty in Artificial Intelligence
Book subtitle proceedings of the Thirty-Fourth Concerence (2018) : August 6-10, 2018, Monterey, California, USA
ISBN (electronic)
  • 978099664319
Event 34th Conference on Uncertainty in Artificial Intelligence
Pages (from-to) 393-402
Publisher Corvallis, Oregon: AUAI Press
Organisations
  • Faculty of Science (FNWI) - Informatics Institute (IVI)
Abstract
Variational inference relies on flexible approximate posterior distributions. Normalizing flows provide a general recipe to construct flexible variational posteriors. We introduce Sylvester normalizing flows, which can 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 Sylvester normalizing flows against planar flows and inverse autoregressive flows and demonstrate that they compare favorably on several datasets.
Document type Conference contribution
Note With supplementary material
Language English
Published at http://auai.org/uai2018/proceedings/papers/156.pdf http://auai.org/uai2018/proceedings/uai2018proceedings.pdf
Other links http://auai.org/uai2018/proceedings/supplements/Supplementary-Paper156.pdf
Downloads
156 (Accepted author manuscript)
Supplementary materials
Permalink to this page
Back