Multiplicative Normalizing Flows for Variational Bayesian Neural Networks
Abstract
We reinterpret multiplicative noise in neural networks as auxiliary random variables that augment the approximate posterior in a variational setting for Bayesian neural networks. We show that through this interpretation it is both efficient and straightforward to improve the approximation by employing normalizing flows while still allowing for local reparametrizations and a tractable lower bound. In experiments we show that with this new approximation we can significantly improve upon classical mean field for Bayesian neural networks on both predictive accuracy as well as predictive uncertainty.
Cite
Text
Louizos and Welling. "Multiplicative Normalizing Flows for Variational Bayesian Neural Networks." International Conference on Machine Learning, 2017.Markdown
[Louizos and Welling. "Multiplicative Normalizing Flows for Variational Bayesian Neural Networks." International Conference on Machine Learning, 2017.](https://mlanthology.org/icml/2017/louizos2017icml-multiplicative/)BibTeX
@inproceedings{louizos2017icml-multiplicative,
title = {{Multiplicative Normalizing Flows for Variational Bayesian Neural Networks}},
author = {Louizos, Christos and Welling, Max},
booktitle = {International Conference on Machine Learning},
year = {2017},
pages = {2218-2227},
volume = {70},
url = {https://mlanthology.org/icml/2017/louizos2017icml-multiplicative/}
}