User: Guest  Login
Title:

Learning Likelihoods with Conditional Normalizing Flows

Document type:
Working Paper
Author(s):
Christina Winkler, Daniel E. Worrall, Emiel Hoogeboom and Max Welling
Abstract:
Normalizing Flows (NFs) are able to model complicated distributions p(y) with strong inter-dimensional correlations and high multimodality by transforming a simple base density p(z) through an invertible neural network under the change of variables formula. Such behavior is desirable in multivariate structured prediction tasks, where handcrafted per-pixel loss-based methods inadequately capture strong correlations between output dimensions. We present a study of conditional normalizing flows (CN...     »
Contracting organization:
University of Amsterdam
Year:
2019
WWW:
https://arxiv.org/pdf/1912.00042.pdf
CC license:
by, http://creativecommons.org/licenses/by/4.0
 BibTeX