Multiplicative normalizing flows for variational Bayesian neural networks
conference paper
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 (Rezende & Mohamed, 2015) while still allowing for local reparametrizations (Kingma et al., 2015) and a tractable lower bound (Ranganath et al., 2015; Maaløe et al., 2016). 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
Topics
TNO Identifier
810219
ISBN
9781510855144
Publisher
International Machine Learning Society (IMLS)
Source title
34th International Conference on Machine Learning, ICML 2017. 6 August 2017 through 11 August 2017
Pages
3480-3489
Files
To receive the publication files, please send an e-mail request to TNO Repository.